[PATCH] git-gui: handle bare repo or missing worktree

public inbox for git@vger.kernel.org 
 help / color / mirror / Atom feed

* [PATCH] git-gui: handle bare repo or missing worktree
@ 2026-04-21 16:28 Shroom Moo
  2026-04-29  6:58 ` Johannes Sixt
  0 siblings, 1 reply; 134+ messages in thread
From: Shroom Moo @ 2026-04-21 16:28 UTC (permalink / raw)
  To: git; +Cc: Johannes Sixt, Shroom Moo

When starting git-gui from a directory that Git recognizes as a valid
repository but the repository is either bare or its working tree is
missing, git-gui previously attempted to execute 'rev-parse
--show-toplevel' without error handling. This caused a fatal Tcl error
("this operation must be run in a work tree") and prevented the user
from opening the repository selection dialog.

Improve the repository setup logic:
- After obtaining the git directory via 'rev-parse --git-dir', check
  whether the repository is bare or if the working tree can be
  successfully located.
- If the repository is unusable as a working tree, display a warning
  and present the "Create/Clone/Open" repository selection dialog.

This makes git-gui robust when launched from a bare repository, from
inside a .git directory without a worktree, or when GIT_DIR points to
an invalid location. No regressions observed in normal working trees.

Signed-off-by: Shroom Moo <egg_mushroomcow@foxmail•com>
---
 git-gui/git-gui.sh | 75 +++++++++++++++++++++++++++++++++++-----------
 1 file changed, 57 insertions(+), 18 deletions(-)

diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh
index 23fe76e498..2e4bc2f226 100755
--- a/git-gui/git-gui.sh
+++ b/git-gui/git-gui.sh
@@ -1123,24 +1123,63 @@ unset argv0dir
 ## repository setup
 
 set picked 0
-if {[catch {
-		set _gitdir $env(GIT_DIR)
-		set _prefix {}
-		}]
-	&& [catch {
-		# beware that from the .git dir this sets _gitdir to .
-		# and _prefix to the empty string
-		set _gitdir [git rev-parse --git-dir]
-		set _prefix [git rev-parse --show-prefix]
-	} err]} {
-	load_config 1
-	apply_config
-	choose_repository::pick
-	if {![file isdirectory $_gitdir]} {
-		exit 1
-	}
-	set picked 1
-}
+# Save temporarily for restoration later
+set _startup_cwd [pwd]
+
+set need_pick 0
+if {[catch {set _gitdir $env(GIT_DIR); set _prefix {}}]} {
+    # GIT_DIR not set in environment, attempt auto-detection
+	# beware that from the .git dir this sets _gitdir to .
+	# and _prefix to the empty string
+    if {[catch {set _gitdir [git rev-parse --git-dir]; set _prefix [git rev-parse --show-prefix]} err]} {
+        # Not inside any Git repository, proceed to repository selection UI
+        set need_pick 1
+    } else {
+        # Inside a Git repository, but need to verify its usability
+        cd [file dirname $_gitdir]
+        set bare 0
+        set worktree_missing 0
+		# Check if this is a bare repository (no working tree)
+        if {![catch {set bare [git rev-parse --is-bare-repository]}]} {
+            if {$bare eq "true"} {
+                set bare 1
+            }
+        }
+		# Check if working tree is present and accessible
+        if {[catch {git rev-parse --show-toplevel}]} {
+            set worktree_missing 1
+        }
+
+        cd $_startup_cwd
+
+        # For bare repos or missing worktrees, warn and guide to selection
+        if {$bare || $worktree_missing} {
+            set msg [mc "The repository at '%s' cannot be opened:" [file normalize $_gitdir]]
+            if {$bare} {
+                append msg "\n\n" [mc "It is a bare repository (no working tree)."]
+            } else {
+                append msg "\n\n" [mc "The working tree appears to be missing or inaccessible."]
+            }
+            append msg "\n\n" [mc "Please select a valid working repository, or create/clone one."]
+            tk_messageBox -icon warning -type ok -title [mc "Repository Unusable"] -message $msg
+
+            set need_pick 1
+        }
+    }
+}
+
+if {$need_pick} {
+    load_config 1
+    apply_config
+    choose_repository::pick
+    if {![file isdirectory $_gitdir]} {
+        exit 1
+    }
+    set picked 1
+}
+
+# Clean up working tree checking temporary variables
+unset -nocomplain _startup_cwd need_pick bare worktree_missing
 
 # Use object format as hash algorithm (either "sha1" or "sha256")
 set hashalgorithm [git rev-parse --show-object-format]
-- 
2.52.0.windows.1


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH] git-gui: handle bare repo or missing worktree
  2026-04-21 16:28 [PATCH] git-gui: handle bare repo or missing worktree Shroom Moo
@ 2026-04-29  6:58 ` Johannes Sixt
  2026-04-29 17:32   ` [PATCH v2 1/1] git-gui: protect rev-parse --show-toplevel call Shroom Moo
  2026-04-29 18:28   ` [PATCH] git-gui: handle bare repo or missing worktree Shroom Moo
  0 siblings, 2 replies; 134+ messages in thread
From: Johannes Sixt @ 2026-04-29  6:58 UTC (permalink / raw)
  To: Shroom Moo; +Cc: git

Am 21.04.26 um 18:28 schrieb Shroom Moo:
> When starting git-gui from a directory that Git recognizes as a valid
> repository but the repository is either bare or its working tree is
> missing, git-gui previously attempted to execute 'rev-parse
> --show-toplevel' without error handling. This caused a fatal Tcl error
> ("this operation must be run in a work tree") and prevented the user
> from opening the repository selection dialog.
> 
> Improve the repository setup logic:
> - After obtaining the git directory via 'rev-parse --git-dir', check
>   whether the repository is bare or if the working tree can be
>   successfully located.
> - If the repository is unusable as a working tree, display a warning
>   and present the "Create/Clone/Open" repository selection dialog.

I consider it harmful to allow the user to create (or clone) a
repository after Git GUI has already determined the existence of a
repository. If the worktree was not found, we should report an error and
exit after the message is dismissed, and not give the option to create a
repository.

> 
> This makes git-gui robust when launched from a bare repository, from
> inside a .git directory without a worktree, or when GIT_DIR points to
> an invalid location. No regressions observed in normal working trees.
> 
> Signed-off-by: Shroom Moo <egg_mushroomcow@foxmail•com>
> ---
>  git-gui/git-gui.sh | 75 +++++++++++++++++++++++++++++++++++-----------
>  1 file changed, 57 insertions(+), 18 deletions(-)
> 
> diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh
> index 23fe76e498..2e4bc2f226 100755
> --- a/git-gui/git-gui.sh
> +++ b/git-gui/git-gui.sh
> @@ -1123,24 +1123,63 @@ unset argv0dir
>  ## repository setup
>  
>  set picked 0
> -if {[catch {
> -		set _gitdir $env(GIT_DIR)
> -		set _prefix {}
> -		}]
> -	&& [catch {
> -		# beware that from the .git dir this sets _gitdir to .
> -		# and _prefix to the empty string
> -		set _gitdir [git rev-parse --git-dir]
> -		set _prefix [git rev-parse --show-prefix]
> -	} err]} {
> -	load_config 1
> -	apply_config
> -	choose_repository::pick
> -	if {![file isdirectory $_gitdir]} {
> -		exit 1
> -	}
> -	set picked 1
> -}
> +# Save temporarily for restoration later
> +set _startup_cwd [pwd]
> +
> +set need_pick 0
> +if {[catch {set _gitdir $env(GIT_DIR); set _prefix {}}]} {
> +    # GIT_DIR not set in environment, attempt auto-detection
> +	# beware that from the .git dir this sets _gitdir to .
> +	# and _prefix to the empty string
> +    if {[catch {set _gitdir [git rev-parse --git-dir]; set _prefix [git rev-parse --show-prefix]} err]} {
> +        # Not inside any Git repository, proceed to repository selection UI
> +        set need_pick 1
> +    } else {
> +        # Inside a Git repository, but need to verify its usability
> +        cd [file dirname $_gitdir]
> +        set bare 0
> +        set worktree_missing 0
> +		# Check if this is a bare repository (no working tree)
> +        if {![catch {set bare [git rev-parse --is-bare-repository]}]} {
> +            if {$bare eq "true"} {
> +                set bare 1
> +            }
> +        }
> +		# Check if working tree is present and accessible
> +        if {[catch {git rev-parse --show-toplevel}]} {
> +            set worktree_missing 1
> +        }
> +
> +        cd $_startup_cwd
> +
> +        # For bare repos or missing worktrees, warn and guide to selection
> +        if {$bare || $worktree_missing} {
> +            set msg [mc "The repository at '%s' cannot be opened:" [file normalize $_gitdir]]
> +            if {$bare} {
> +                append msg "\n\n" [mc "It is a bare repository (no working tree)."]
> +            } else {
> +                append msg "\n\n" [mc "The working tree appears to be missing or inaccessible."]
> +            }
> +            append msg "\n\n" [mc "Please select a valid working repository, or create/clone one."]
> +            tk_messageBox -icon warning -type ok -title [mc "Repository Unusable"] -message $msg
> +
> +            set need_pick 1
> +        }
> +    }
> +}
> +
> +if {$need_pick} {
> +    load_config 1
> +    apply_config
> +    choose_repository::pick
> +    if {![file isdirectory $_gitdir]} {
> +        exit 1
> +    }
> +    set picked 1
> +}
> +
> +# Clean up working tree checking temporary variables
> +unset -nocomplain _startup_cwd need_pick bare worktree_missing

This adds quite a lot of code with failure cases, but after this point
we already have some quite extensive error diagnosis, too. Except that
the call to git rev-parse --show-toplevel is not protected.

If we do not invoke the Create/Clone/Open dialog after a repository has
already been discovered, would it not be sufficient to just add the
protection around the --show-toplevel call? Or is there something else
missing in the existing error paths?

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v2 1/1] git-gui: protect rev-parse --show-toplevel call
  2026-04-29  6:58 ` Johannes Sixt
@ 2026-04-29 17:32   ` Shroom Moo
  2026-04-29 20:14     ` Mark Levedahl
  2026-04-30 10:02     ` [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir Shroom Moo
  2026-04-29 18:28   ` [PATCH] git-gui: handle bare repo or missing worktree Shroom Moo
  1 sibling, 2 replies; 134+ messages in thread
From: Shroom Moo @ 2026-04-29 17:32 UTC (permalink / raw)
  To: git; +Cc: j6t, Shroom Moo

When starting git-gui from a directory that is a bare repository or
where the working tree is missing, git-gui previously executed
'rev-parse --show-toplevel' without error handling. This caused a
fatal Tcl error ("this operation must be run in a work tree").

Wrap the call in a catch to prevent the fatal error. The existing
error paths after this call already handle bare repos and missing
worktrees appropriately.

Signed-off-by: Shroom Moo <egg_mushroomcow@foxmail•com>
---
 git-gui/git-gui.sh | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh
index 23fe76e498..aee37685e1 100755
--- a/git-gui/git-gui.sh
+++ b/git-gui/git-gui.sh
@@ -1169,7 +1169,9 @@ if {![file isdirectory $_gitdir]} {
 load_config 0
 apply_config
 
-set _gitworktree [git rev-parse --show-toplevel]
+if {[catch {set _gitworktree [git rev-parse --show-toplevel]}]} {
+    set _gitworktree {}
+}
 
 if {$_prefix ne {}} {
 	if {$_gitworktree eq {}} {
-- 
2.52.0.windows.1


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 1/1] git-gui: protect rev-parse --show-toplevel call
  2026-04-29 17:32   ` [PATCH v2 1/1] git-gui: protect rev-parse --show-toplevel call Shroom Moo
@ 2026-04-29 20:14     ` Mark Levedahl
  2026-04-30 10:02     ` [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir Shroom Moo
  1 sibling, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-04-29 20:14 UTC (permalink / raw)
  To: Shroom Moo, git; +Cc: j6t

On 4/29/26 1:32 PM, Shroom Moo wrote:
> When starting git-gui from a directory that is a bare repository or
> where the working tree is missing, git-gui previously executed
> 'rev-parse --show-toplevel' without error handling. This caused a
> fatal Tcl error ("this operation must be run in a work tree").
> 
> Wrap the call in a catch to prevent the fatal error. The existing
> error paths after this call already handle bare repos and missing
> worktrees appropriately.
> 
> Signed-off-by: Shroom Moo <egg_mushroomcow@foxmail•com>
> ---
>  git-gui/git-gui.sh | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh
> index 23fe76e498..aee37685e1 100755
> --- a/git-gui/git-gui.sh
> +++ b/git-gui/git-gui.sh
> @@ -1169,7 +1169,9 @@ if {![file isdirectory $_gitdir]} {
>  load_config 0
>  apply_config
>  
> -set _gitworktree [git rev-parse --show-toplevel]
> +if {[catch {set _gitworktree [git rev-parse --show-toplevel]}]} {
> +    set _gitworktree {}
> +}
>  
>  if {$_prefix ne {}} {
>  	if {$_gitworktree eq {}} {

Unfortunately, this allows starting git-gui inside the separate gitdir created by

	git clone --separate-git-dir=/some/where/else ...

There is no hint where the workdir is, but git recognizes the repository is not bare:

	git rev-parse --is-bare-repository  ==> false
	git rev-parse --is-inside-git-dir   ==> true
	git rev-parse --is-inside-work-tree ==> false
	git rev-parse --show-toplevel       ==> fatal: must be run in a work tree
	git worktree list                   ==> absolute path to gitdir (not a worktree)

As git-gui has no protection against modifying what is effectively a bare repository,
allowing git-gui to run in this directory is dangerous, or possibly just very confusing.

git refuses to work in this gitdir: "git status" run in the above gitdir gives: "fatal:
this operation must be run in a work tree."

The simplest safe thing is to catch the error and abort with a more useful message than
currently provided. Or perhaps, check git rev-parse --is-inside-git-dir and abort, and do
so before trying --show-toplevel.


Mark


^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir
  2026-04-29 17:32   ` [PATCH v2 1/1] git-gui: protect rev-parse --show-toplevel call Shroom Moo
  2026-04-29 20:14     ` Mark Levedahl
@ 2026-04-30 10:02     ` Shroom Moo
  2026-04-30 16:18       ` Mark Levedahl
  2026-05-01 10:54       ` [PATCH v4 " Shroom Moo
  1 sibling, 2 replies; 134+ messages in thread
From: Shroom Moo @ 2026-04-30 10:02 UTC (permalink / raw)
  To: git; +Cc: j6t, mlevedahl, Shroom Moo

When git-gui is started from a directory that Git recognizes as a
valid repository but the working tree is not accessible (e.g., a
separated gitdir created by `git clone --separate-git-dir`, a bare
repository, or a case where the worktree directory was removed),
it previously called `rev-parse --show-toplevel` without error
handling, causing a fatal Tcl error ("this operation must be run
in a work tree").

Wrap the call in a `catch` and handle the failure as follows:

- For bare repositories, keep `_gitworktree` empty so that the
  existing `is_bare` check shows "Cannot use bare repository" and
  exits.  No behavioral change.

- For non‑bare repositories, try to locate the worktree from the
  parent directory using `git -C $parent rev-parse --show-toplevel`.
  If the parent is a valid worktree, change to it; this covers the
  legitimate case of starting git-gui from within the .git
  subdirectory of a normal working tree.

- If the parent directory is not a worktree, refuse to start with
  a clear error message.  This prevents dangerous operations in a
  separated gitdir, where ordinary Git commands like `git status`
  would themselves refuse to run.

The approach intentionally avoids two pitfalls:

  - Testing `--is-inside-git-dir` before calling `--show-toplevel`
    would break the normal use case of starting git-gui from within
    a .git subdirectory (where --show-toplevel would succeed).

  - A simple “non‑bare” check after a failed --show-toplevel would
    reject a normal repository whose worktree was only temporarily
    removed.

The chosen method keeps the original behavior for bare repositories
and for regular working trees, fixes the crash, and properly blocks
separated gitdirs without a reachable worktree.

Signed-off-by: Shroom Moo <egg_mushroomcow@foxmail•com>
---
 git-gui/git-gui.sh | 23 ++++++++++++++++++++++-
 1 file changed, 22 insertions(+), 1 deletion(-)

diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh
index 23fe76e498..2392282df3 100755
--- a/git-gui/git-gui.sh
+++ b/git-gui/git-gui.sh
@@ -1169,7 +1169,28 @@ if {![file isdirectory $_gitdir]} {
 load_config 0
 apply_config
 
-set _gitworktree [git rev-parse --show-toplevel]
+if {[catch {set _gitworktree [git rev-parse --show-toplevel]}]} {
+	# For bare repositories, use the existing error handling
+	if {![catch {set bare [git rev-parse --is-bare-repository]}] && $bare eq {true}} {
+		set _gitworktree {}
+	} else {
+		# Non-bare: try to find the worktree from the parent directory
+		set parent [file dirname [pwd]]
+		# Cannot go higher than the root directory; leave _gitworktree empty
+		if {[file normalize $parent] eq [file normalize [pwd]]} {
+			# Already at the filesystem root; let existing paths cope
+			set _gitworktree {}
+		} elseif {![catch {
+			set _gitworktree [git -C $parent rev-parse --show-toplevel]
+		}]} {
+			cd $parent
+		} else {
+			catch {wm withdraw .}
+			error_popup [mc "Cannot start git-gui from inside the Git directory."]
+			exit 1
+		}
+	}
+}
 
 if {$_prefix ne {}} {
 	if {$_gitworktree eq {}} {
-- 
2.52.0.windows.1


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir
  2026-04-30 10:02     ` [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir Shroom Moo
@ 2026-04-30 16:18       ` Mark Levedahl
  2026-05-01 10:22         ` [PATCH v3 1/1] git-gui: handle missing worktree and separated Shroom Moo
  2026-05-01 13:13         ` [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir Johannes Sixt
  2026-05-01 10:54       ` [PATCH v4 " Shroom Moo
  1 sibling, 2 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-04-30 16:18 UTC (permalink / raw)
  To: Shroom Moo, git; +Cc: j6t



On 4/30/26 6:02 AM, Shroom Moo wrote:
> When git-gui is started from a directory that Git recognizes as a
> valid repository but the working tree is not accessible (e.g., a
> separated gitdir created by `git clone --separate-git-dir`, a bare
> repository, or a case where the worktree directory was removed),
> it previously called `rev-parse --show-toplevel` without error
> handling, causing a fatal Tcl error ("this operation must be run
> in a work tree").
>
> Wrap the call in a `catch` and handle the failure as follows:
>
> - For bare repositories, keep `_gitworktree` empty so that the
>   existing `is_bare` check shows "Cannot use bare repository" and
>   exits.  No behavioral change.
>
> - For non‑bare repositories, try to locate the worktree from the
>   parent directory using `git -C $parent rev-parse --show-toplevel`.
>   If the parent is a valid worktree, change to it; this covers the
>   legitimate case of starting git-gui from within the .git
>   subdirectory of a normal working tree.
>
> - If the parent directory is not a worktree, refuse to start with
>   a clear error message.  This prevents dangerous operations in a
>   separated gitdir, where ordinary Git commands like `git status`
>   would themselves refuse to run.
>
> The approach intentionally avoids two pitfalls:
>
>   - Testing `--is-inside-git-dir` before calling `--show-toplevel`
>     would break the normal use case of starting git-gui from within
>     a .git subdirectory (where --show-toplevel would succeed).
>
>   - A simple “non‑bare” check after a failed --show-toplevel would
>     reject a normal repository whose worktree was only temporarily
>     removed.
>
> The chosen method keeps the original behavior for bare repositories
> and for regular working trees, fixes the crash, and properly blocks
> separated gitdirs without a reachable worktree.
>
> Signed-off-by: Shroom Moo <egg_mushroomcow@foxmail•com>
> ---
>  git-gui/git-gui.sh | 23 ++++++++++++++++++++++-
>  1 file changed, 22 insertions(+), 1 deletion(-)
>
> diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh
> index 23fe76e498..2392282df3 100755
> --- a/git-gui/git-gui.sh
> +++ b/git-gui/git-gui.sh
> @@ -1169,7 +1169,28 @@ if {![file isdirectory $_gitdir]} {
>  load_config 0
>  apply_config
>  
> -set _gitworktree [git rev-parse --show-toplevel]
> +if {[catch {set _gitworktree [git rev-parse --show-toplevel]}]} {
> +	# For bare repositories, use the existing error handling
> +	if {![catch {set bare [git rev-parse --is-bare-repository]}] && $bare eq {true}} {
> +		set _gitworktree {}
> +	} else {
> +		# Non-bare: try to find the worktree from the parent directory
> +		set parent [file dirname [pwd]]
> +		# Cannot go higher than the root directory; leave _gitworktree empty
> +		if {[file normalize $parent] eq [file normalize [pwd]]} {
> +			# Already at the filesystem root; let existing paths cope
> +			set _gitworktree {}
> +		} elseif {![catch {
> +			set _gitworktree [git -C $parent rev-parse --show-toplevel]
> +		}]} {
> +			cd $parent
> +		} else {
> +			catch {wm withdraw .}
> +			error_popup [mc "Cannot start git-gui from inside the Git directory."]
> +			exit 1
> +		}
> +	}
> +}
>  
>  if {$_prefix ne {}} {
>  	if {$_gitworktree eq {}} {

A bare repository can be contained in a workdir / worktree pointing at a different gitdir:
the logic above can thus a workdir that doesn't use the gitdir where git-gui was started.
The bigger issue is that a gitdir can support multiple checked-out directories with no
one-to-one mapping and no clear idea of which of those a user may have intended.

So, I believe the correct fix is to test "rev-parse --is-inside-git-dir, and if so throw a
clear error message and exit. This will give the user something to start with to solve the
problem of why they started git-gui in a gitdir, and not in a worktree.

Mark


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v3 1/1] git-gui: handle missing worktree and separated
  2026-04-30 16:18       ` Mark Levedahl
@ 2026-05-01 10:22         ` Shroom Moo
  2026-05-01 13:13         ` [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir Johannes Sixt
  1 sibling, 0 replies; 134+ messages in thread
From: Shroom Moo @ 2026-05-01 10:22 UTC (permalink / raw)
  To: mlevedahl, git; +Cc: Shroom Moo, j6t

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1255 bytes --]

Hi Mark,

Thanks for catching the multi‑worktree ambiguity. The parent‑directory
fallback can surely accidentally pick the wrong worktree. As a 
reminder, the current code deliberately supports starting git-gui from 
within a regular repository's .git directory.

The comment says:
# beware that from the .git dir this sets _gitdir to .
# and _prefix to the empty string

In that case, _gitdir is ".", _prefix is empty, and the later logic
falls back to using [file dirname $_gitdir] as the worktree. A blanket
"if --is-inside-git-dir then exit" would make that case useless.

I'll send a v4 that first checks --is-bare-repository (preserving the
original bare‑repo error), then checks --is-inside-git-dir and refuses
if inside a gitdir. This accepts the .git‑startup limitation in
exchange for safety, and keeps the bare‑repo message unchanged.

Two alternatives still exist if a different trade‑off is preferred: 
- Only check --is-inside-git-dir (simpler, but makes the
bare‑repo error_popup useless).
- After --is-inside-git-dir, consult git worktree list and
switch to the single worktree if unambiguous (keeps .git‑startup
but adds complexity and a runtime dependency).

Shroom

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir
  2026-04-30 16:18       ` Mark Levedahl
  2026-05-01 10:22         ` [PATCH v3 1/1] git-gui: handle missing worktree and separated Shroom Moo
@ 2026-05-01 13:13         ` Johannes Sixt
  2026-05-01 16:42           ` Mark Levedahl
  1 sibling, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-01 13:13 UTC (permalink / raw)
  To: Mark Levedahl, Shroom Moo; +Cc: git

Am 30.04.26 um 18:18 schrieb Mark Levedahl:
> 
> 
> On 4/30/26 6:02 AM, Shroom Moo wrote:
>> When git-gui is started from a directory that Git recognizes as a
>> valid repository but the working tree is not accessible (e.g., a
>> separated gitdir created by `git clone --separate-git-dir`, a bare
>> repository, or a case where the worktree directory was removed),
>> it previously called `rev-parse --show-toplevel` without error
>> handling, causing a fatal Tcl error ("this operation must be run
>> in a work tree").
>>
>> Wrap the call in a `catch` and handle the failure as follows:
>>
>> - For bare repositories, keep `_gitworktree` empty so that the
>>   existing `is_bare` check shows "Cannot use bare repository" and
>>   exits.  No behavioral change.
>>
>> - For non‑bare repositories, try to locate the worktree from the
>>   parent directory using `git -C $parent rev-parse --show-toplevel`.
>>   If the parent is a valid worktree, change to it; this covers the
>>   legitimate case of starting git-gui from within the .git
>>   subdirectory of a normal working tree.
>>
>> - If the parent directory is not a worktree, refuse to start with
>>   a clear error message.  This prevents dangerous operations in a
>>   separated gitdir, where ordinary Git commands like `git status`
>>   would themselves refuse to run.
>>
>> The approach intentionally avoids two pitfalls:
>>
>>   - Testing `--is-inside-git-dir` before calling `--show-toplevel`
>>     would break the normal use case of starting git-gui from within
>>     a .git subdirectory (where --show-toplevel would succeed).
>>
>>   - A simple “non‑bare” check after a failed --show-toplevel would
>>     reject a normal repository whose worktree was only temporarily
>>     removed.
>>
>> The chosen method keeps the original behavior for bare repositories
>> and for regular working trees, fixes the crash, and properly blocks
>> separated gitdirs without a reachable worktree.
>>
>> Signed-off-by: Shroom Moo <egg_mushroomcow@foxmail•com>
>> ---
>>  git-gui/git-gui.sh | 23 ++++++++++++++++++++++-
>>  1 file changed, 22 insertions(+), 1 deletion(-)
>>
>> diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh
>> index 23fe76e498..2392282df3 100755
>> --- a/git-gui/git-gui.sh
>> +++ b/git-gui/git-gui.sh
>> @@ -1169,7 +1169,28 @@ if {![file isdirectory $_gitdir]} {
>>  load_config 0
>>  apply_config
>>  
>> -set _gitworktree [git rev-parse --show-toplevel]
>> +if {[catch {set _gitworktree [git rev-parse --show-toplevel]}]} {
>> +	# For bare repositories, use the existing error handling
>> +	if {![catch {set bare [git rev-parse --is-bare-repository]}] && $bare eq {true}} {
>> +		set _gitworktree {}
>> +	} else {
>> +		# Non-bare: try to find the worktree from the parent directory
>> +		set parent [file dirname [pwd]]
>> +		# Cannot go higher than the root directory; leave _gitworktree empty
>> +		if {[file normalize $parent] eq [file normalize [pwd]]} {
>> +			# Already at the filesystem root; let existing paths cope
>> +			set _gitworktree {}
>> +		} elseif {![catch {
>> +			set _gitworktree [git -C $parent rev-parse --show-toplevel]
>> +		}]} {
>> +			cd $parent
>> +		} else {
>> +			catch {wm withdraw .}
>> +			error_popup [mc "Cannot start git-gui from inside the Git directory."]
>> +			exit 1
>> +		}
>> +	}
>> +}
>>  
>>  if {$_prefix ne {}} {
>>  	if {$_gitworktree eq {}} {
> 
> A bare repository can be contained in a workdir / worktree pointing at a different gitdir:
> the logic above can thus a workdir that doesn't use the gitdir where git-gui was started.
> The bigger issue is that a gitdir can support multiple checked-out directories with no
> one-to-one mapping and no clear idea of which of those a user may have intended.
> 
> So, I believe the correct fix is to test "rev-parse --is-inside-git-dir, and if so throw a
> clear error message and exit. This will give the user something to start with to solve the
> problem of why they started git-gui in a gitdir, and not in a worktree.
We have quite a bit of code that attempts to make Git GUI work from the
.git directory and also in bare repositories.

87cd09f43e56 ("git-gui: work from the .git dir", 2010-01-23) made the
first step. The original code just used the $_gitdir as the working
directory. However, at that time we did not have alternate worktrees,
and the old code, when used today, does not work in a `git
worktree`-created worktree. Later, the `git rev-parse --show-toplevel`
call came with 38ec8d3e2652 ("git-gui: correct assignment of work-tree",
2010-10-20). However, it also changes the fall-back code slightly, so
that running Git GUI from the .git directory would not work the same way
as before and takes the .git directory as the work tree (because in the
.git directory --show-cdup is not "..", but empty).

I think we need to restructure the existing flow a bit and not just fix
a single spot in the code. I suggest this order of operation:

1. Handle the bare repository case. If not enabled, fail. Otherwise, we
can work with an empty $_gitworktree.

2. Collect --show-toplevel into $_gitworktree.

2a. If this failed: If --is-inside-git-dir is true, and the last
$_gitdir directory component is exactly ".git", take the parent
repository as $_gitworktree. Otherwise, fail.

3. Handle all the other edge cases, if any, with the so determined
$_gitworktree. (I didn't think through, yet, what needs to be done.)

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir
  2026-05-01 13:13         ` [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir Johannes Sixt
@ 2026-05-01 16:42           ` Mark Levedahl
  2026-05-02 21:51             ` Mark Levedahl
  0 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-01 16:42 UTC (permalink / raw)
  To: Johannes Sixt, Shroom Moo; +Cc: git

On 5/1/26 9:13 AM, Johannes Sixt wrote:
> Am 30.04.26 um 18:18 schrieb Mark Levedahl:
>>
>> On 4/30/26 6:02 AM, Shroom Moo wrote:
>> We have quite a bit of code that attempts to make Git GUI work from the
> .git directory and also in bare repositories.
>
> 87cd09f43e56 ("git-gui: work from the .git dir", 2010-01-23) made the
> first step. The original code just used the $_gitdir as the working
> directory. However, at that time we did not have alternate worktrees,
> and the old code, when used today, does not work in a `git
> worktree`-created worktree. Later, the `git rev-parse --show-toplevel`
> call came with 38ec8d3e2652 ("git-gui: correct assignment of work-tree",
> 2010-10-20). However, it also changes the fall-back code slightly, so
> that running Git GUI from the .git directory would not work the same way
> as before and takes the .git directory as the work tree (because in the
> .git directory --show-cdup is not "..", but empty).
>
> I think we need to restructure the existing flow a bit and not just fix
> a single spot in the code. I suggest this order of operation:
>
> 1. Handle the bare repository case. If not enabled, fail. Otherwise, we
> can work with an empty $_gitworktree.
>
> 2. Collect --show-toplevel into $_gitworktree.
>
> 2a. If this failed: If --is-inside-git-dir is true, and the last
> $_gitdir directory component is exactly ".git", take the parent
> repository as $_gitworktree. Otherwise, fail.
>
> 3. Handle all the other edge cases, if any, with the so determined
> $_gitworktree. (I didn't think through, yet, what needs to be done.)
>
> -- Hannes
>

I found one horrid edge case: 

Start git-gui in a gitdir not embedded in a worktree, with core.bare=false as there are
one or more gitfile and/or symlinked worktrees elsewhere.
- current git-gui aborts with an uncaught error. Good.
- git-gui with the wrapped --show-toplevel call finds no worktree to switch to, so runs in
the gitdir allowing commits of the gitdir items.
-  I just added and  committed the *file* refs/heads/master to branch master in such a gitdir.

git-gui's normal gui must be started ONLY if rev-parse --is-inside-work-tree is true. (The
blame view invoked by gitk in theory could be allowed to run in a bare repository
read-only mode.).

For read/write mode:
if --is-inside-git-dir == true at startup, we must abort, or find a valid worktree and
switch to that.

My personal preference is for git-gui to abort:
   I started git-gui where it cannot run. 
   My error. Let me learn and fix that.

Alternatively, ask me what to do:
    e.g., prepare a dialog after looking at git-worktree list, and the parent dir IFF this
dir is named .git, telling me of my mistake and offering me one or more worktrees to
switch to.

But please, don't just switch to another directory without asking. This is just
encouraging me to make careless errors.

Mark


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir
  2026-05-01 16:42           ` Mark Levedahl
@ 2026-05-02 21:51             ` Mark Levedahl
  2026-05-03  8:53               ` Johannes Sixt
  0 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-02 21:51 UTC (permalink / raw)
  To: Johannes Sixt, Shroom Moo; +Cc: git



On 5/1/26 12:42 PM, Mark Levedahl wrote:
> On 5/1/26 9:13 AM, Johannes Sixt wrote:
>> Am 30.04.26 um 18:18 schrieb Mark Levedahl:
>>> On 4/30/26 6:02 AM, Shroom Moo wrote:
>>> We have quite a bit of code that attempts to make Git GUI work from the
>> .git directory and also in bare repositories.
>>
>> 87cd09f43e56 ("git-gui: work from the .git dir", 2010-01-23) made the
>> first step. The original code just used the $_gitdir as the working
>> directory. However, at that time we did not have alternate worktrees,
>> and the old code, when used today, does not work in a `git
>> worktree`-created worktree. Later, the `git rev-parse --show-toplevel`
>> call came with 38ec8d3e2652 ("git-gui: correct assignment of work-tree",
>> 2010-10-20). However, it also changes the fall-back code slightly, so
>> that running Git GUI from the .git directory would not work the same way
>> as before and takes the .git directory as the work tree (because in the
>> .git directory --show-cdup is not "..", but empty).
>>
>> I think we need to restructure the existing flow a bit and not just fix
>> a single spot in the code. I suggest this order of operation:
>>
>> 1. Handle the bare repository case. If not enabled, fail. Otherwise, we
>> can work with an empty $_gitworktree.
>>
>> 2. Collect --show-toplevel into $_gitworktree.
>>
>> 2a. If this failed: If --is-inside-git-dir is true, and the last
>> $_gitdir directory component is exactly ".git", take the parent
>> repository as $_gitworktree. Otherwise, fail.
>>
>> 3. Handle all the other edge cases, if any, with the so determined
>> $_gitworktree. (I didn't think through, yet, what needs to be done.)
>>
>> -- Hannes
>>
> I found one horrid edge case: 
>
> Start git-gui in a gitdir not embedded in a worktree, with core.bare=false as there are
> one or more gitfile and/or symlinked worktrees elsewhere.
> - current git-gui aborts with an uncaught error. Good.
> - git-gui with the wrapped --show-toplevel call finds no worktree to switch to, so runs in
> the gitdir allowing commits of the gitdir items.
> -  I just added and  committed the *file* refs/heads/master to branch master in such a gitdir.
>
> git-gui's normal gui must be started ONLY if rev-parse --is-inside-work-tree is true. (The
> blame view invoked by gitk in theory could be allowed to run in a bare repository
> read-only mode.).
>
> For read/write mode:
> if --is-inside-git-dir == true at startup, we must abort, or find a valid worktree and
> switch to that.
>
> My personal preference is for git-gui to abort:
>    I started git-gui where it cannot run. 
>    My error. Let me learn and fix that.
>
> Alternatively, ask me what to do:
>     e.g., prepare a dialog after looking at git-worktree list, and the parent dir IFF this
> dir is named .git, telling me of my mistake and offering me one or more worktrees to
> switch to.
>
> But please, don't just switch to another directory without asking. This is just
> encouraging me to make careless errors.
>
> Mark
>
I dug a bit more into the startup logic, and I think I better understand rework that is
needed.

Two basic problems I see here are beyond the question of if (and when) git-gui should try
to locate a worktree:

- git gui blame in a gitdir was apparently broken by the git repo commit 2d92ab32fd
("rev-parse: make --show-toplevel without a worktree an error", 2019-11-19). Prior to that
commit, git gui would stay in the startup directory enabling only features that cannot
modify the repository, and gitk could bring this view up in a gitdir. This doesn't work
right now.

- git-gui's logic includes a conceptual error embodied in proc is_bare: is_bare uses $(git
rev-parse --is-bare-repository) but what we need is $(git rev-parse --is-inside-git-dir),
and these are not synonyms.
It does not matter whether a worktree exists that points at the gitdir, and as discussed
before, main worktrees can easily exist that we cannot locate from the gitdir. At best,
is_bare is a guess.
So, is_bare should be replaced by is_inside_gitdir, and we should also have
is_inside_worktree. These are mutually exclusive, though both can be false.

My current idea of an improved startup flow enables features only at the end:

1) If not in a gitdir or worktree, 
    1a)   if git gui's subcommand is not 'gui', abort with an error (citool, browser, or
blame invocations carry information specific to a gitdir/worktree). 
    1b)   otherwise, invoke repository_chooser, which either aborts, or changes directory
to a worktree.

-- we are now in a gitdir or a worktree, and this may or may not be the startup directory.

2) Look at the combination of git gui subcommand and directory type (worktree / gitdir) to
decide to continue.
    2a) blame / browser are ok in either directory type.
    2b) citool requires a specific worktree, which should have been the initial startup
directory. Abort if not.
    2b) gui requires a worktree. Abort if not (my recommendation), or offer to find (or
automatically find) a worktree.

3) Change directory to the top level of of the directory_type (git rev-parse knows
toplevel of a worktree, different code is needed for a gitdir).

4) Enable features based upon subcommand and directory type.

There are 12 combinations of initial directory type (gitdir, worktree, neither) and
subcommand (gui, blame, browser, citool) to consider, with a lot of duplicated code
amongst the 12 cases. So, obviously, steps 1 and 2 can be convolved in many ways that are
different than what I wrote above.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir
  2026-05-02 21:51             ` Mark Levedahl
@ 2026-05-03  8:53               ` Johannes Sixt
  2026-05-04 15:13                 ` Mark Levedahl
  0 siblings, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-03  8:53 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: git, Shroom Moo

Am 02.05.26 um 23:51 schrieb Mark Levedahl:
> Two basic problems I see here are beyond the question of if (and when) git-gui should try
> to locate a worktree:
> 
> - git gui blame in a gitdir was apparently broken by the git repo commit 2d92ab32fd
> ("rev-parse: make --show-toplevel without a worktree an error", 2019-11-19). Prior to that
> commit, git gui would stay in the startup directory enabling only features that cannot
> modify the repository, and gitk could bring this view up in a gitdir. This doesn't work
> right now.

True, `blame` used to work in bare repositories, but is broken now.

> - git-gui's logic includes a conceptual error embodied in proc is_bare: is_bare uses $(git
> rev-parse --is-bare-repository) but what we need is $(git rev-parse --is-inside-git-dir),
> and these are not synonyms.
> It does not matter whether a worktree exists that points at the gitdir, and as discussed
> before, main worktrees can easily exist that we cannot locate from the gitdir. At best,
> is_bare is a guess.
> So, is_bare should be replaced by is_inside_gitdir, and we should also have
> is_inside_worktree. These are mutually exclusive, though both can be false.

I would not call the use of --is-bar-repository instead of
--is-inside-git-dir an error, just a choice that has been made. In
particular, when the startup directory is named '.git' and is not marked
as bare, then its parent directory can very reasonably be taken as its
worktree. (That's how things worked before --show-toplevel was used.) If
the check is for --is-inside-git-dir, this treatment would be ruled out
early.


> My current idea of an improved startup flow enables features only at the end:
> 
> 1) If not in a gitdir or worktree, 
>     1a)   if git gui's subcommand is not 'gui', abort with an error (citool, browser, or
> blame invocations carry information specific to a gitdir/worktree). 
>     1b)   otherwise, invoke repository_chooser, which either aborts, or changes directory
> to a worktree.
> 
> -- we are now in a gitdir or a worktree, and this may or may not be the startup directory.
> 
> 2) Look at the combination of git gui subcommand and directory type (worktree / gitdir) to
> decide to continue.
>     2a) blame / browser are ok in either directory type.
>     2b) citool requires a specific worktree, which should have been the initial startup
> directory. Abort if not.
>     2b) gui requires a worktree. Abort if not (my recommendation), or offer to find (or
> automatically find) a worktree.
> 
> 3) Change directory to the top level of of the directory_type (git rev-parse knows
> toplevel of a worktree, different code is needed for a gitdir).
> 
> 4) Enable features based upon subcommand and directory type.
> 
> There are 12 combinations of initial directory type (gitdir, worktree, neither) and
> subcommand (gui, blame, browser, citool) to consider, with a lot of duplicated code
> amongst the 12 cases. So, obviously, steps 1 and 2 can be convolved in many ways that are
> different than what I wrote above.
So true.

But perhaps there is a simpler solution: Let's present an error if
--show-toplevel fails except in the case where the startup directory is
named '.git' (and is a valid Git repository) and is not bare (then the
worktree is the parent). I insist in this exception, because this
use-case was considered important in the past (87cd09f43e56 "git-gui:
work from the .git dir", 2010-01-23).

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir
  2026-05-03  8:53               ` Johannes Sixt
@ 2026-05-04 15:13                 ` Mark Levedahl
  2026-05-05  3:40                   ` Mark Levedahl
  2026-05-06  7:32                   ` Johannes Sixt
  0 siblings, 2 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-04 15:13 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: git, Shroom Moo

On 5/3/26 4:53 AM, Johannes Sixt wrote:
> I would not call the use of --is-bar-repository instead of
> --is-inside-git-dir an error, just a choice that has been made. In
> particular, when the startup directory is named '.git' and is not marked
> as bare, then its parent directory can very reasonably be taken as its
> worktree. (That's how things worked before --show-toplevel was used.) If
> the check is for --is-inside-git-dir, this treatment would be ruled out
> early.
>
Whether being in a gitdir is ok, or a worktree required, is of fundamental importance and
is not explicitly checked now. This is my issue. (Whether the repo is bare, or embedded in
a worktree, is relevant only when automatically fixing a user error.)
> But perhaps there is a simpler solution: Let's present an error if
> --show-toplevel fails except in the case where the startup directory is
> named '.git' (and is a valid Git repository) and is not bare (then the
> worktree is the parent). I insist in this exception, because this
> use-case was considered important in the past (87cd09f43e56 "git-gui:
> work from the .git dir", 2010-01-23).
>
> -- Hannes
>

This would not fix gitk's blame / browse from a gitdir, and I don't really see a one or
two line fix as being adequate.

git-gui sets GIT_WORK_TREE and GIT_DIR at startup. GIT_DIR passes my simple tests, but
mishandles GIT_WORK_TREE.

I expect these two invocations to be equivalent, both starting git-gui in the worktree
'/some/path':

    GIT_WORK_TREE=/some/path git gui
    git -C /some/path gui

But, the GIT_WORK_TREE approach:
    works as I expect ONLY when the current directory is a valid worktree
    when started from a gitdir, uses that gitdir in conjunction with the requested worktree
    when started from an uncontrolled directory, shows the repository picker.

The git -C approach is indifferent to the current directory, of course.

GIT_WORK_TREE enters much too late in the process, and rather should handled first:
    if GIT_WORK_TREE is in the environment, cd to that first. Throw an error if that
directory is not a valid worktree.

I don't actually understand the use case of defining GIT_DIR or GIT_WORK_TREE to git gui,
and I wonder what other bugs are lurking... maybe the better approach is to just abort if
GIT_DIR or GIT_WORK_TREE are defined?

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir
  2026-05-04 15:13                 ` Mark Levedahl
@ 2026-05-05  3:40                   ` Mark Levedahl
  2026-05-06  7:32                   ` Johannes Sixt
  1 sibling, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-05  3:40 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: git, Shroom Moo

>
> On 5/3/26 4:53 AM, Johannes Sixt wrote:
>> But perhaps there is a simpler solution: Let's present an error if
>> --show-toplevel fails except in the case where the startup directory is
>> named '.git' (and is a valid Git repository) and is not bare (then the
>> worktree is the parent). I insist in this exception, because this
>> use-case was considered important in the past (87cd09f43e56 "git-gui:
>> work from the .git dir", 2010-01-23).
>>
>> -- Hannes
>>

I've restructured startup code in line with what I suggested before, allowing operation in
a worktree or a gitdir, and with various combinations of GIT_DIR and GIT_WORK_TREE
environment variables set. Unfortunately, git-gui's blame and browser commands simply do
not now work without a valid worktree. The error(s) are not obvious to me, and bisecting
requires git version 2.24 or earlier to remove 2d92ab32fd ("rev-parse: make
--show-toplevel without a worktree an error", 2019-11-19) to even start: many gcc and
git/git-gui version compatibility issues are certain to arise. I won't be doing this.

So, I believe your suggestion above is the best path, leaving behind dead code that
purports to support operation from a gitdir but does not.

Mark



^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir
  2026-05-04 15:13                 ` Mark Levedahl
  2026-05-05  3:40                   ` Mark Levedahl
@ 2026-05-06  7:32                   ` Johannes Sixt
  2026-05-06 11:27                     ` Mark Levedahl
  1 sibling, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-06  7:32 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: git, Shroom Moo

Am 04.05.26 um 17:13 schrieb Mark Levedahl:
> On 5/3/26 4:53 AM, Johannes Sixt wrote:
>> I would not call the use of --is-bar-repository instead of
>> --is-inside-git-dir an error, just a choice that has been made. In
>> particular, when the startup directory is named '.git' and is not marked
>> as bare, then its parent directory can very reasonably be taken as its
>> worktree. (That's how things worked before --show-toplevel was used.) If
>> the check is for --is-inside-git-dir, this treatment would be ruled out
>> early.
>>
> Whether being in a gitdir is ok, or a worktree required, is of fundamental importance and
> is not explicitly checked now. This is my issue. (Whether the repo is bare, or embedded in
> a worktree, is relevant only when automatically fixing a user error.)

I don't quite follow what you a trying to say here.

>> But perhaps there is a simpler solution: Let's present an error if
>> --show-toplevel fails except in the case where the startup directory is
>> named '.git' (and is a valid Git repository) and is not bare (then the
>> worktree is the parent). I insist in this exception, because this
>> use-case was considered important in the past (87cd09f43e56 "git-gui:
>> work from the .git dir", 2010-01-23).
>>
>> -- Hannes
>>
> 
> This would not fix gitk's blame / browse from a gitdir, and I don't really see a one or
> two line fix as being adequate.
> 
> git-gui sets GIT_WORK_TREE and GIT_DIR at startup. GIT_DIR passes my simple tests, but
> mishandles GIT_WORK_TREE.
> 
> I expect these two invocations to be equivalent, both starting git-gui in the worktree
> '/some/path':
> 
>     GIT_WORK_TREE=/some/path git gui
>     git -C /some/path gui
> 
> But, the GIT_WORK_TREE approach:
>     works as I expect ONLY when the current directory is a valid worktree
>     when started from a gitdir, uses that gitdir in conjunction with the requested worktree
>     when started from an uncontrolled directory, shows the repository picker.

The important aspect here isn't about the worktree, but whether a gitdir
can be determined for the current directory. All three observations make
total sense.

That said, setting GIT_WORK_TREE without also setting GIT_DIR is
undefined and need not be considered further.

> The git -C approach is indifferent to the current directory, of course.
> 
> GIT_WORK_TREE enters much too late in the process, and rather should handled first:
>     if GIT_WORK_TREE is in the environment, cd to that first. Throw an error if that
> directory is not a valid worktree.

As I said, GIT_WORK_TREE without GIT_DIR is an invalid use-case. For
this reason, the first thing to do is find the database, and from there
work out the worktree. In the most common use-case it is the current
directory.

> I don't actually understand the use case of defining GIT_DIR or GIT_WORK_TREE to git gui,
> and I wonder what other bugs are lurking... maybe the better approach is to just abort if
> GIT_DIR or GIT_WORK_TREE are defined?
I lean towards setting GIT_DIR always. This is necessary, because Git
GUI can be run from a subdirectory of the worktree, and then changes
directory to the top-level. It must be ensured that the same GIT_DIR is
used that was detected from the subdirectory.

Now that the current directory is at the top-level of the worktree, we
could just not set GIT_WORK_TREE at all, provided that setting GIT_DIR
without GIT_WORK_TREE is a valid use-case for Git. I am not yet sure
about that.

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir
  2026-05-06  7:32                   ` Johannes Sixt
@ 2026-05-06 11:27                     ` Mark Levedahl
  2026-05-06 12:57                       ` Johannes Sixt
  0 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-06 11:27 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: git, Shroom Moo

On 5/6/26 3:32 AM, Johannes Sixt wrote:
>> Whether being in a gitdir is ok, or a worktree required, is of fundamental importance and
>> is not explicitly checked now. This is my issue. (Whether the repo is bare, or embedded in
>> a worktree, is relevant only when automatically fixing a user error.)
> I don't quite follow what you a trying to say here.
I played a bit more:

A git repository (gitdir) can have config.bare true | false | not set
git rev-parse --is-bare-repository tells you that whatever gitdir is discovered from the
current directory has core.bare==true. This happens whether the call is from inside the
gitdir, or in the parent dir of a gitdir named '.git', or in a directory containing a
symlink or a gitfile link to the gitdir. This call never tells you what directory you are
actually in.

git rev-parse --is-inside-work-tree gives:
    true - the call is made from a directory that is suported/supportable as a worktree of
a gitdir.
    false - the call is made from inside a gitdir, or from a directory linked to a to a
gitdir with core.bare == true.
    and error is thrown if no gitdir is discovered.

I find --is-inside-work-tree a much better call to make early in setup. 
    true - full git-gui is ok, 
    false - blame/browser is ok (gitdir might have core.bare true)
    error - no gitdir found, the repository picker should be called.

So, the only need to test if the repo is marked bare is when looking for a possible
worktree when git-gui was started inside the gitdir, or started in a directory linked to
said gitdir, or GIT_DIR in the environment points to said gitdir: I consider all of this a
user (or configuration) error, and there are many possible causes to explore to give
useful feedback to the user.

But, there are many ways to code this. I started down a path of using
--is-inside-worktree, but in the end there are still a lot of corner cases to find.

>>> But perhaps there is a simpler solution: Let's present an error if
>>> --show-toplevel fails except in the case where the startup directory is
>>> named '.git' (and is a valid Git repository) and is not bare (then the
>>> worktree is the parent). I insist in this exception, because this
>>> use-case was considered important in the past (87cd09f43e56 "git-gui:
>>> work from the .git dir", 2010-01-23).
>>>
>>> -- Hannes
>>>
>> This would not fix gitk's blame / browse from a gitdir, and I don't really see a one or
>> two line fix as being adequate.

As you mentioned elsewhere, the problem on browser/blame is that _gitworktree is empty
when no worktree is found, so GIT_WORK_TREE is exported to the environment as an empty
variable. This cause is in a commit from 12 years ago:

    3decb8e0ac ("git-gui: tolerate major version changes when comparing the git version",
2014-05-17)

Prior to that commit and if not using git v1.7.x, an alternate branch of code not using
git rev-parse was used for worktree discovery, and that code set _gitworktree = _gitdir
when in a gitdir. The alternate code was removed more recently as it was unreachable from
non-ancient git versions.

The fix is to set _gitworktree to _gitdir before exporting GIT_WORK_TREE, or to just not
export an empty GIT_WORK_TREE. Obviously, having GIT_WORK_TREE = GIT_DIR is asking for
trouble, but perhaps is ok as git-gui is running in a read-only mode for browse/blame. My
limited testing shows this works.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir
  2026-05-06 11:27                     ` Mark Levedahl
@ 2026-05-06 12:57                       ` Johannes Sixt
  2026-05-06 14:05                         ` Mark Levedahl
  0 siblings, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-06 12:57 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: git, Shroom Moo

Am 06.05.26 um 13:27 schrieb Mark Levedahl:
> A git repository (gitdir) can have config.bare true | false | not set
> git rev-parse --is-bare-repository tells you that whatever gitdir is discovered from the
> current directory has core.bare==true. This happens whether the call is from inside the
> gitdir, or in the parent dir of a gitdir named '.git', or in a directory containing a
> symlink or a gitfile link to the gitdir. This call never tells you what directory you are
> actually in.

OK. But how does "find out which directory we are in" come into play
here? If we find a bare repository, we do not need a worktree. If we are
in a non-bare repository, we can find the worktree with `rev-parse
--show-toplevel`.

> 
> git rev-parse --is-inside-work-tree gives:
>     true - the call is made from a directory that is suported/supportable as a worktree of
> a gitdir.
>     false - the call is made from inside a gitdir, or from a directory linked to a to a
> gitdir with core.bare == true.
>     and error is thrown if no gitdir is discovered.
> 
> I find --is-inside-work-tree a much better call to make early in setup. 
>     true - full git-gui is ok, 
>     false - blame/browser is ok (gitdir might have core.bare true)
>     error - no gitdir found, the repository picker should be called.

But we would still make an exception for the case that $PWD is a
non-bare repository named ".git", because then, by Git GUI's definition,
its parent is the corresponding worktree.

> So, the only need to test if the repo is marked bare is when looking for a possible
> worktree when git-gui was started inside the gitdir, or started in a directory linked to
> said gitdir, or GIT_DIR in the environment points to said gitdir: I consider all of this a
> user (or configuration) error, and there are many possible causes to explore to give
> useful feedback to the user.

How does this scheme work when the user starts `git gui blame` in a bare
repository that does not have a worktree? Would this not produce an
error because no worktree was found?

> As you mentioned elsewhere, the problem on browser/blame is that _gitworktree is empty
> when no worktree is found, so GIT_WORK_TREE is exported to the environment as an empty
> variable. This cause is in a commit from 12 years ago:
> 
>     3decb8e0ac ("git-gui: tolerate major version changes when comparing the git version",
> 2014-05-17)

I don't think that this commit very relevant. The problem is in `git
branch --show-current` (and probably other git command variants) that
want to turn an empty $GIT_WORK_TREE into an absolute path even in cases
where no worktree is needed. I haven't tried to figure out which commit
(in the Git repository) started to do this.

> The fix is to set _gitworktree to _gitdir before exporting GIT_WORK_TREE, or to just not
> export an empty GIT_WORK_TREE. Obviously, having GIT_WORK_TREE = GIT_DIR is asking for
> trouble, but perhaps is ok as git-gui is running in a read-only mode for browse/blame. My
> limited testing shows this works.
Good to know. My preference is to not set GIT_WORK_TREE at all provided
that setting GIT_DIR without GIT_WORK_TREE is a use-case supported by Git.

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir
  2026-05-06 12:57                       ` Johannes Sixt
@ 2026-05-06 14:05                         ` Mark Levedahl
  2026-05-07  5:09                           ` Mark Levedahl
  0 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-06 14:05 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: git, Shroom Moo



On 5/6/26 8:57 AM, Johannes Sixt wrote:
> Am 06.05.26 um 13:27 schrieb Mark Levedahl:
>> A git repository (gitdir) can have config.bare true | false | not set
>> git rev-parse --is-bare-repository tells you that whatever gitdir is discovered from the
>> current directory has core.bare==true. This happens whether the call is from inside the
>> gitdir, or in the parent dir of a gitdir named '.git', or in a directory containing a
>> symlink or a gitfile link to the gitdir. This call never tells you what directory you are
>> actually in.
> OK. But how does "find out which directory we are in" come into play
> here? If we find a bare repository, we do not need a worktree. If we are
> in a non-bare repository, we can find the worktree with `rev-parse
> --show-toplevel`.
>
>> git rev-parse --is-inside-work-tree gives:
>>     true - the call is made from a directory that is suported/supportable as a worktree of
>> a gitdir.
>>     false - the call is made from inside a gitdir, or from a directory linked to a to a
>> gitdir with core.bare == true.
>>     and error is thrown if no gitdir is discovered.
>>
>> I find --is-inside-work-tree a much better call to make early in setup. 
>>     true - full git-gui is ok, 
>>     false - blame/browser is ok (gitdir might have core.bare true)
>>     error - no gitdir found, the repository picker should be called.
> But we would still make an exception for the case that $PWD is a
> non-bare repository named ".git", because then, by Git GUI's definition,
> its parent is the corresponding worktree.

I find the organization using rev-parse --is-inside-work-tree easier to reason about, and
if I were writing this from scratch, I would do it that way. But, you have one or more
patches in progress, if this idea is useful there great, otherwise, drop it.

>> As you mentioned elsewhere, the problem on browser/blame is that _gitworktree is empty
>> when no worktree is found, so GIT_WORK_TREE is exported to the environment as an empty
>> variable. This cause is in a commit from 12 years ago:
>>
>>     3decb8e0ac ("git-gui: tolerate major version changes when comparing the git version",
>> 2014-05-17)
> I don't think that this commit very relevant. The problem is in `git
> branch --show-current` (and probably other git command variants) that
> want to turn an empty $GIT_WORK_TREE into an absolute path even in cases
> where no worktree is needed. I haven't tried to figure out which commit
> (in the Git repository) started to do this.

Reverting that commit in any way has nothing to do with fixing this problem now. But,
detecting breakage at that commit is what lead me to discover the problem was _gitworktree
== {} and GIT_WORK_TREE="". As you say, browser/blame may well not have broken until a
more recent commit to git itself. I cannot say, even the 2019 git prior to rev parse
--top-level being taught to error out will not build on my computer due to incompatibilities.

>> The fix is to set _gitworktree to _gitdir before exporting GIT_WORK_TREE, or to just not
>> export an empty GIT_WORK_TREE. Obviously, having GIT_WORK_TREE = GIT_DIR is asking for
>> trouble, but perhaps is ok as git-gui is running in a read-only mode for browse/blame. My
>> limited testing shows this works.
> Good to know. My preference is to not set GIT_WORK_TREE at all provided
> that setting GIT_DIR without GIT_WORK_TREE is a use-case supported by Git.
>
>

I just confirmed that git-gui modified to export GIT_DIR only, not GIT_WORK_TREE, and
actually to make sure GIT_WORK_TREE is not in env, has blame/browser working correctly in
a gitdir with no worktree.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir
  2026-05-06 14:05                         ` Mark Levedahl
@ 2026-05-07  5:09                           ` Mark Levedahl
  0 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-07  5:09 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: git, Shroom Moo

On 5/6/26 10:05 AM, Mark Levedahl wrote:
>
> On 5/6/26 8:57 AM, Johannes Sixt wrote:
>> Am 06.05.26 um 13:27 schrieb Mark Levedahl:
>>> A git repository (gitdir) can have config.bare true | false | not set
>>> git rev-parse --is-bare-repository tells you that whatever gitdir is discovered from the
>>> current directory has core.bare==true. This happens whether the call is from inside the
>>> gitdir, or in the parent dir of a gitdir named '.git', or in a directory containing a
>>> symlink or a gitfile link to the gitdir. This call never tells you what directory you are
>>> actually in.
>> OK. But how does "find out which directory we are in" come into play
>> here? If we find a bare repository, we do not need a worktree. If we are
>> in a non-bare repository, we can find the worktree with `rev-parse
>> --show-toplevel`.
The influence of config variable core.bare in a git repository breaks the design I
contemplated using rev-parse --is-inside-work-tree. core.bare is a strange tri-state item,
usually only set if the repository is explicitly bare. With core.bare undefined, rev-parse
can give different answers inside and outside a gitdir. For instance
      rev-parse --is-bare-repository gives true inside a repository, and false outside,
--is-inside-work-tree is also influenced by core.bare, so results cannot be interpreted as
I initially thought.

Anyway, after some experiments, the minimal sufficient tests I find for blame|browser and
gui|citool are different:

- blame|browser need a readable gitdir, and do not need a worktree. The current directory
can be inside the gitdir, core.bare is irrelevant.

    a sufficient startup test:

        git rev-parse --git-dir returns the name of a gitdir, rather than throwing an error.
               core.bare and being inside/outside of the gitdir do affect error / no error.
                (the return value may be a relative path, so can depend upon current
directory)

    only GIT_DIR should be exported. GIT_WORK_TREE should not be in env.

- gui|citool, require a gitdir, a worktree outside the gitdir, and core.bare is not true

    a sufficient and minimal startup test:

        git rev-parse --show-toplevel returns the worktree root, rather than throwing an
error.
                core.bare = true in the gitdir will cause error in the above.
                being inside the gitdir causes error in the above.

        But, starting in a .git directory supporting a worktree is allowed by the fallback
logic

        git rev-parse --git-dir returns '.', (current dir is the root of the gitdir),
rather than throwing an error.
        the current directory is named ".git", meaning the parent dir is this gitdir's
worktree if core.bare is not true
        git -C .. rev-parse --show-toplevel  returns the worktree root: (will throw an
error if core.bare is true)

    the current directory should be changed to the worktree root.
    GIT_DIR and GIT_WORK_TREE should be exported to env.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v4 1/1] git-gui: handle missing worktree and separated gitdir
  2026-04-30 10:02     ` [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir Shroom Moo
  2026-04-30 16:18       ` Mark Levedahl
@ 2026-05-01 10:54       ` Shroom Moo
  2026-05-04 14:59         ` [PATCH v5 1/1] git-gui: restructure repository startup Shroom Moo
  1 sibling, 1 reply; 134+ messages in thread
From: Shroom Moo @ 2026-05-01 10:54 UTC (permalink / raw)
  To: git; +Cc: j6t, mlevedahl, Shroom Moo

When git-gui is started from a directory that Git recognizes as a
valid repository but the working tree is not accessible (e.g., a
separated gitdir created by `git clone --separate-git-dir`, a bare
repository, or a case where the worktree directory was removed),
it previously called `rev-parse --show-toplevel` without error
handling, causing a fatal Tcl error ("this operation must be run
in a work tree").

Wrap the call in a `catch` and handle the failure as follows:

- For bare repositories, keep `_gitworktree` empty so that the
  existing `is_bare` check shows "Cannot use bare repository" and
  exits.  No behavioral change.

- If we are inside the gitdir (i.e., `--is-inside-git-dir` is
  `true`), refuse to start with a clear message: "Cannot start
  git-gui from inside the Git directory. Please run git-gui from a
  working tree."  This prevents dangerous operations in a separated
  gitdir or when starting from a .git subdirectory, where ordinary
  Git commands like `git status` also refuse to run.

- Otherwise, leave `_gitworktree` empty so that the existing error
  paths (e.g., missing worktree) can display their own diagnostics.

This fix intentionally drops the previous ability to start git-gui
from within a regular repository's .git subdirectory.  That behavior
was never reliable (Git itself forbids operations inside .git) and
broke completely in the multi‑worktree case where the parent
directory is not guaranteed to be the intended worktree.  The error
message guides the user to start git-gui from a worktree instead.

Signed-off-by: Shroom Moo <egg_mushroomcow@foxmail•com>
---
 git-gui/git-gui.sh | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh
index 23fe76e498..9e848aa26b 100755
--- a/git-gui/git-gui.sh
+++ b/git-gui/git-gui.sh
@@ -1169,7 +1169,17 @@ if {![file isdirectory $_gitdir]} {
 load_config 0
 apply_config
 
-set _gitworktree [git rev-parse --show-toplevel]
+if {[catch {set _gitworktree [git rev-parse --show-toplevel]}]} {
+    if {![catch {set bare [git rev-parse --is-bare-repository]}] && $bare eq {true}} {
+        set _gitworktree {}
+    } elseif {![catch {set inside [git rev-parse --is-inside-git-dir]}] && $inside eq {true}} {
+        catch {wm withdraw .}
+        error_popup [mc "Cannot start git-gui from inside the Git directory\nPlease run git-gui from a working tree"]
+        exit 1
+    } else {
+        set _gitworktree {}
+    }
+}
 
 if {$_prefix ne {}} {
 	if {$_gitworktree eq {}} {
-- 
2.52.0.windows.1


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* [PATCH v5 1/1] git-gui: restructure repository startup
  2026-05-01 10:54       ` [PATCH v4 " Shroom Moo
@ 2026-05-04 14:59         ` Shroom Moo
  2026-05-06  7:15           ` Johannes Sixt
                             ` (2 more replies)
  0 siblings, 3 replies; 134+ messages in thread
From: Shroom Moo @ 2026-05-04 14:59 UTC (permalink / raw)
  To: git; +Cc: Johannes Sixt, Mark Levedahl, Shroom Moo

When git-gui is started inside a .git directory of a non-bare
repository, it should treat the parent directory as the worktree,
as it did before commit 2d92ab32fd (rev-parse: make --show-toplevel
without a worktree an error, 2019-11-19).  However, a bare repository
or a separated gitdir without a worktree must be rejected early.

Protect the previously unguarded calls to `git rev-parse
--show-object-format` and `--show-toplevel`.  Restructure the startup
sequence to:

- Check for a bare repository right after loading the config.  If the
  repository is bare and the current subcommand does not allow bare
  repos (e.g. normal commit mode), show "Cannot use bare repository"
  and exit.

- When `rev-parse --show-toplevel` fails and the repository is
  non-bare, the gitdir path ends with ".git", and we are inside that
  gitdir, use the parent directory as the worktree.  This preserves
  the ability to start git-gui from within a regular repository’s
  .git directory, which was intentionally supported since 87cd09f43e56
  (git-gui: work from the .git dir, 2010-01-23).

- Otherwise, show a descriptive error and exit.

- Wrap `rev-parse --show-object-format` in a catch to avoid a crash
  when the repository configuration is broken (e.g. core.worktree
  pointing to an invalid path).

Also removes the old `_prefix`‑based fallback that computed a relative
path to the worktree top from a subdirectory, and the unconditional
`[file dirname $_gitdir]` guess.  Both are unnecessary now that
`rev‑parse --show‑toplevel` directly provides the absolute top‑level
path and we can `cd` to it.  The guess is further unsafe in
multi‑worktree setups, where a gitdir may have more than one worktree.
The only remaining fallback is the explicit “.git directory” rule for
non‑bare repositories, which mirrors the historical behaviour.

This fixes the fatal Tcl error when the working tree is missing, while
keeping the .git startup feature and avoiding any automatic directory
switching that could be dangerous in multi‑worktree setups.

Signed-off-by: Shroom Moo <egg_mushroomcow@foxmail•com>
---
 git-gui/git-gui.sh | 72 +++++++++++++++++++++++++++++-----------------
 1 file changed, 45 insertions(+), 27 deletions(-)

diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh
index 23fe76e498..c06d85b8d9 100755
--- a/git-gui/git-gui.sh
+++ b/git-gui/git-gui.sh
@@ -1129,7 +1129,8 @@ if {[catch {
 		}]
 	&& [catch {
 		# beware that from the .git dir this sets _gitdir to .
-		# and _prefix to the empty string
+		# and _prefix to the empty string; this is handled by
+		# the startup safety checks below
 		set _gitdir [git rev-parse --git-dir]
 		set _prefix [git rev-parse --show-prefix]
 	} err]} {
@@ -1142,8 +1143,20 @@ if {[catch {
 	set picked 1
 }
 
+if {![file isdirectory $_gitdir]} {
+	catch {wm withdraw .}
+	error_popup [strcat 
+		[mc "Git directory not found:"] "\n\n$_gitdir\n\n" \
+		[mc "Please ensure GIT_DIR points to a valid Git repository"]]
+	exit 1
+}
+
 # Use object format as hash algorithm (either "sha1" or "sha256")
-set hashalgorithm [git rev-parse --show-object-format]
+if {[catch {set hashalgorithm [git rev-parse --show-object-format]} err]} {
+	catch {wm withdraw .}
+	error_popup [strcat [mc "Failed to determine hash algorithm:"] "\n\n$err"]
+	exit 1
+}
 if {$hashalgorithm eq "sha1"} {
 	set hashlength 40
 } elseif {$hashalgorithm eq "sha256"} {
@@ -1160,46 +1173,50 @@ if {$_gitdir eq "."} {
 	set _gitdir [pwd]
 }
 
-if {![file isdirectory $_gitdir]} {
-	catch {wm withdraw .}
-	error_popup [strcat [mc "Git directory not found:"] "\n\n$_gitdir"]
-	exit 1
-}
 # _gitdir exists, so try loading the config
 load_config 0
 apply_config
 
-set _gitworktree [git rev-parse --show-toplevel]
+# Handle bare repository early: if not allowed, abort
+if {[is_bare] && ![is_enabled bare]} {
+	catch {wm withdraw .}
+	error_popup [strcat [mc "Cannot use bare repository:"] "\n\n" [file normalize $_gitdir]]
+	exit 1
+}
 
-if {$_prefix ne {}} {
-	if {$_gitworktree eq {}} {
-		regsub -all {[^/]+/} $_prefix ../ cdup
-	} else {
-		set cdup $_gitworktree
-	}
-	if {[catch {cd $cdup} err]} {
-		catch {wm withdraw .}
-		error_popup [strcat [mc "Cannot move to top of working directory:"] "\n\n$err"]
-		exit 1
+# Determine the working tree
+if {[is_bare] && [is_enabled bare]} {
+	set _gitworktree {}
+} else {
+	if {[catch {set _gitworktree [git rev-parse --show-toplevel]} err]} {
+		# If we are inside a .git directory of a non-bare repo,
+		# the worktree is the parent directory
+		set inside_gitdir 0
+		catch {set inside_gitdir [git rev-parse --is-inside-git-dir]}
+		if {![is_bare] && $inside_gitdir eq {true} && [file tail [file normalize $_gitdir]] eq {.git}} {
+			set _gitworktree [file normalize [file dirname $_gitdir]]
+		} else {
+			catch {wm withdraw .}
+			error_popup [strcat [mc "Cannot determine working tree:"] "\n\n$err"]
+			exit 1
+		}
 	}
-	set _gitworktree [pwd]
-	unset cdup
-} elseif {![is_enabled bare]} {
-	if {[is_bare]} {
+
+	if {$_gitworktree eq {}} {
 		catch {wm withdraw .}
-		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n$_gitdir"]
+		error_popup [mc "Cannot determine working tree (unexpected empty result)"]
 		exit 1
 	}
-	if {$_gitworktree eq {}} {
-		set _gitworktree [file dirname $_gitdir]
-	}
+
 	if {[catch {cd $_gitworktree} err]} {
 		catch {wm withdraw .}
-		error_popup [strcat [mc "No working directory"] " $_gitworktree:\n\n$err"]
+		error_popup [strcat [mc "Cannot move to working directory:"] "\n\n$err"]
 		exit 1
 	}
 	set _gitworktree [pwd]
 }
+
+# Derive a human-readable repository name
 set _reponame [file split [file normalize $_gitdir]]
 if {[lindex $_reponame end] eq {.git}} {
 	set _reponame [lindex $_reponame end-1]
@@ -1207,6 +1224,7 @@ if {[lindex $_reponame end] eq {.git}} {
 	set _reponame [lindex $_reponame end]
 }
 
+# Export the final paths
 set env(GIT_DIR) $_gitdir
 set env(GIT_WORK_TREE) $_gitworktree
 
-- 
2.52.0.windows.1


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v5 1/1] git-gui: restructure repository startup
  2026-05-04 14:59         ` [PATCH v5 1/1] git-gui: restructure repository startup Shroom Moo
@ 2026-05-06  7:15           ` Johannes Sixt
  2026-05-06 20:27           ` [PATCH v6 0/3] git-gui: robustify startup and fix environment handling Shroom Moo
       [not found]           ` <20260506202751.3294-1-egg_mushroomcow@foxmail.com>
  2 siblings, 0 replies; 134+ messages in thread
From: Johannes Sixt @ 2026-05-06  7:15 UTC (permalink / raw)
  To: Shroom Moo; +Cc: Mark Levedahl, git

Am 04.05.26 um 16:59 schrieb Shroom Moo:
> When git-gui is started inside a .git directory of a non-bare
> repository, it should treat the parent directory as the worktree,
> as it did before commit 2d92ab32fd (rev-parse: make --show-toplevel
> without a worktree an error, 2019-11-19).  However, a bare repository
> or a separated gitdir without a worktree must be rejected early.
> 
> Protect the previously unguarded calls to `git rev-parse
> --show-object-format` and `--show-toplevel`.  Restructure the startup
> sequence to:
> 
> - Check for a bare repository right after loading the config.  If the
>   repository is bare and the current subcommand does not allow bare
>   repos (e.g. normal commit mode), show "Cannot use bare repository"
>   and exit.
> 
> - When `rev-parse --show-toplevel` fails and the repository is
>   non-bare, the gitdir path ends with ".git", and we are inside that
>   gitdir, use the parent directory as the worktree.  This preserves
>   the ability to start git-gui from within a regular repository’s
>   .git directory, which was intentionally supported since 87cd09f43e56
>   (git-gui: work from the .git dir, 2010-01-23).
> 
> - Otherwise, show a descriptive error and exit.

Very good. This does things in the right order, IMO.

> 
> - Wrap `rev-parse --show-object-format` in a catch to avoid a crash
>   when the repository configuration is broken (e.g. core.worktree
>   pointing to an invalid path).

Nice catch. This could be moved into its own patch. But it is acceptable
in this patch as it loosely fits the topic.

> 
> Also removes the old `_prefix`‑based fallback that computed a relative
> path to the worktree top from a subdirectory, and the unconditional
> `[file dirname $_gitdir]` guess.  Both are unnecessary now that
> `rev‑parse --show‑toplevel` directly provides the absolute top‑level
> path and we can `cd` to it.  The guess is further unsafe in
> multi‑worktree setups, where a gitdir may have more than one worktree.
> The only remaining fallback is the explicit “.git directory” rule for
> non‑bare repositories, which mirrors the historical behaviour.

Nice cleanup.

> 
> This fixes the fatal Tcl error when the working tree is missing, while
> keeping the .git startup feature and avoiding any automatic directory
> switching that could be dangerous in multi‑worktree setups.

Good!

However, this doesn't fix `git gui blame HEAD file` in a bare
repository, because `git branch --show-current` fails with an empty
GIT_WORK_TREE value. Fixing this needs to consider whether to set
GIT_DIR and GIT_WORK_TREE at all, as alluded to by Mark in a near-by
message. This is a separate topic.

> 
> Signed-off-by: Shroom Moo <egg_mushroomcow@foxmail•com>
> ---
>  git-gui/git-gui.sh | 72 +++++++++++++++++++++++++++++-----------------
>  1 file changed, 45 insertions(+), 27 deletions(-)
> 
> diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh
> index 23fe76e498..c06d85b8d9 100755
> --- a/git-gui/git-gui.sh
> +++ b/git-gui/git-gui.sh
> @@ -1129,7 +1129,8 @@ if {[catch {
>  		}]
>  	&& [catch {
>  		# beware that from the .git dir this sets _gitdir to .
> -		# and _prefix to the empty string
> +		# and _prefix to the empty string; this is handled by
> +		# the startup safety checks below
>  		set _gitdir [git rev-parse --git-dir]
>  		set _prefix [git rev-parse --show-prefix]
>  	} err]} {
> @@ -1142,8 +1143,20 @@ if {[catch {
>  	set picked 1
>  }
>  
> +if {![file isdirectory $_gitdir]} {
> +	catch {wm withdraw .}
> +	error_popup [strcat 
> +		[mc "Git directory not found:"] "\n\n$_gitdir\n\n" \
> +		[mc "Please ensure GIT_DIR points to a valid Git repository"]]
> +	exit 1
> +}
> +
>  # Use object format as hash algorithm (either "sha1" or "sha256")
> -set hashalgorithm [git rev-parse --show-object-format]
> +if {[catch {set hashalgorithm [git rev-parse --show-object-format]} err]} {
> +	catch {wm withdraw .}
> +	error_popup [strcat [mc "Failed to determine hash algorithm:"] "\n\n$err"]
> +	exit 1
> +}
>  if {$hashalgorithm eq "sha1"} {
>  	set hashlength 40
>  } elseif {$hashalgorithm eq "sha256"} {
> @@ -1160,46 +1173,50 @@ if {$_gitdir eq "."} {
>  	set _gitdir [pwd]
>  }
>  
> -if {![file isdirectory $_gitdir]} {
> -	catch {wm withdraw .}
> -	error_popup [strcat [mc "Git directory not found:"] "\n\n$_gitdir"]
> -	exit 1
> -}
>  # _gitdir exists, so try loading the config
>  load_config 0
>  apply_config
>  
> -set _gitworktree [git rev-parse --show-toplevel]
> +# Handle bare repository early: if not allowed, abort
> +if {[is_bare] && ![is_enabled bare]} {
> +	catch {wm withdraw .}
> +	error_popup [strcat [mc "Cannot use bare repository:"] "\n\n" [file normalize $_gitdir]]
> +	exit 1
> +}
>  
> -if {$_prefix ne {}} {
> -	if {$_gitworktree eq {}} {
> -		regsub -all {[^/]+/} $_prefix ../ cdup
> -	} else {
> -		set cdup $_gitworktree
> -	}
> -	if {[catch {cd $cdup} err]} {
> -		catch {wm withdraw .}
> -		error_popup [strcat [mc "Cannot move to top of working directory:"] "\n\n$err"]
> -		exit 1
> +# Determine the working tree
> +if {[is_bare] && [is_enabled bare]} {
> +	set _gitworktree {}

I strongly suggest to collapse this branch and the previous 'if
{[is_bare] && ![is_enabled bare]}' into a single 'if {[is_bare]}',
because then in the else-branch below we can be sure not to have a bare
repository.

> +} else {
> +	if {[catch {set _gitworktree [git rev-parse --show-toplevel]} err]} {
> +		# If we are inside a .git directory of a non-bare repo,
> +		# the worktree is the parent directory
> +		set inside_gitdir 0
> +		catch {set inside_gitdir [git rev-parse --is-inside-git-dir]}
> +		if {![is_bare] && $inside_gitdir eq {true} && [file tail [file normalize $_gitdir]] eq {.git}} {

Do we need to 'file normalize' before taking the 'file tail'? If not,
then the line would be shorter.

> +			set _gitworktree [file normalize [file dirname $_gitdir]]
> +		} else {
> +			catch {wm withdraw .}
> +			error_popup [strcat [mc "Cannot determine working tree:"] "\n\n$err"]
> +			exit 1
> +		}
>  	}
> -	set _gitworktree [pwd]
> -	unset cdup
> -} elseif {![is_enabled bare]} {
> -	if {[is_bare]} {
> +
> +	if {$_gitworktree eq {}} {
>  		catch {wm withdraw .}
> -		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n$_gitdir"]
> +		error_popup [mc "Cannot determine working tree (unexpected empty result)"]
>  		exit 1
>  	}
> -	if {$_gitworktree eq {}} {
> -		set _gitworktree [file dirname $_gitdir]
> -	}
> +
>  	if {[catch {cd $_gitworktree} err]} {
>  		catch {wm withdraw .}
> -		error_popup [strcat [mc "No working directory"] " $_gitworktree:\n\n$err"]
> +		error_popup [strcat [mc "Cannot move to working directory:"] "\n\n$err"]
>  		exit 1
>  	}
>  	set _gitworktree [pwd]
>  }
> +
> +# Derive a human-readable repository name
>  set _reponame [file split [file normalize $_gitdir]]
>  if {[lindex $_reponame end] eq {.git}} {
>  	set _reponame [lindex $_reponame end-1]
> @@ -1207,6 +1224,7 @@ if {[lindex $_reponame end] eq {.git}} {
>  	set _reponame [lindex $_reponame end]
>  }
>  
> +# Export the final paths
>  set env(GIT_DIR) $_gitdir
>  set env(GIT_WORK_TREE) $_gitworktree
>  

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v6 0/3] git-gui: robustify startup and fix environment handling
  2026-05-04 14:59         ` [PATCH v5 1/1] git-gui: restructure repository startup Shroom Moo
  2026-05-06  7:15           ` Johannes Sixt
@ 2026-05-06 20:27           ` Shroom Moo
  2026-05-09 13:37             ` [PATCH v7 " Shroom Moo
       [not found]             ` <20260509133756.1367-1-egg_mushroomcow@foxmail.com>
       [not found]           ` <20260506202751.3294-1-egg_mushroomcow@foxmail.com>
  2 siblings, 2 replies; 134+ messages in thread
From: Shroom Moo @ 2026-05-06 20:27 UTC (permalink / raw)
  To: git; +Cc: Johannes Sixt, Mark Levedahl, Shroom Moo

This series addresses the startup crash introduced by Git commit
"2d92ab32fd ("rev-parse: make --show-toplevel without a worktree an
error", 2019-11-19)", which causes `git gui` to die with a Tcl error
when a worktree is missing (e.g. inside a .git directory without a
working tree, or in a bare repository).

Additionally, it resolves two historically inconsistent behaviours:
- The "Visualize ... History" menu items were enabled in bare
  repositories but triggered Tcl errors due to the missing worktree.
- `GIT_WORK_TREE` and `GIT_DIR` environment variables were not
  respected early enough, so `GIT_WORK_TREE=/some/path git gui`
  often ignored the explicit worktree and brought up the repository
  picker, and an exported empty `GIT_WORK_TREE` confused commands
  like `git branch --show-current` in bare repositories.

Shroom Moo (3):
  git-gui: restructure repository startup
  git-gui: disable gitk visualization when no worktree available
  git-gui: handle GIT_DIR and GIT_WORK_TREE early

 git-gui/git-gui.sh | 170 ++++++++++++++++++++++++++++++++++-----------
 1 file changed, 128 insertions(+), 42 deletions(-)

-- 
2.52.0.windows.1


^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v7 0/3] git-gui: robustify startup and fix environment handling
  2026-05-06 20:27           ` [PATCH v6 0/3] git-gui: robustify startup and fix environment handling Shroom Moo
@ 2026-05-09 13:37             ` Shroom Moo
  2026-05-14 14:28               ` Mark Levedahl
       [not found]             ` <20260509133756.1367-1-egg_mushroomcow@foxmail.com>
  1 sibling, 1 reply; 134+ messages in thread
From: Shroom Moo @ 2026-05-09 13:37 UTC (permalink / raw)
  To: git; +Cc: Johannes Sixt, Mark Levedahl, Aina Boot, Shroom Moo

This series addresses the startup crash introduced by Git commit
"2d92ab32fd ("rev-parse: make --show-toplevel without a worktree an
error", 2019-11-19)", which causes `git gui` to die with a Tcl error
when a worktree is missing (e.g. inside a .git directory without a
working tree, or in a bare repository).

Additionally, it resolves two historically inconsistent behaviours:
- The "Visualize ... History" menu items were enabled in bare
  repositories but triggered Tcl errors due to the missing worktree.
- `GIT_WORK_TREE` and `GIT_DIR` environment variables were not
  respected early enough, so `GIT_WORK_TREE=/some/path git gui`
  often ignored the explicit worktree and brought up the repository
  picker, and an exported empty `GIT_WORK_TREE` confused commands
  like `git branch --show-current` in bare repositories.

Shroom Moo (3):
  git-gui: restructure repository startup
  git-gui: disable gitk visualization when no worktree available
  git-gui: handle GIT_DIR and GIT_WORK_TREE early

 git-gui/git-gui.sh | 187 ++++++++++++++++++++++++++++++++++-----------
 1 file changed, 141 insertions(+), 46 deletions(-)

-- 
2.52.0.windows.1


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v7 0/3] git-gui: robustify startup and fix environment handling
  2026-05-09 13:37             ` [PATCH v7 " Shroom Moo
@ 2026-05-14 14:28               ` Mark Levedahl
  2026-05-14 14:33                 ` [PATCH v1 00/11] Improve git gui operation without a worktree Mark Levedahl
  0 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-14 14:28 UTC (permalink / raw)
  To: Shroom Moo, git; +Cc: Johannes Sixt, Aina Boot

On 5/9/26 9:37 AM, Shroom Moo wrote:
> Shroom Moo (3):
>   git-gui: restructure repository startup
>   git-gui: disable gitk visualization when no worktree available
>   git-gui: handle GIT_DIR and GIT_WORK_TREE early
>
>
After careful consideration, I find starting off by fixing what is broken in git-gui about
using a bare-repository, and letting git core handle GIT_DIR and GIT_WORK_TREE, leads to a
much more complete and different solution. A patch series (attempting to) do so will follow.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v1 00/11] Improve git gui operation without a worktree
  2026-05-14 14:28               ` Mark Levedahl
@ 2026-05-14 14:33                 ` Mark Levedahl
  2026-05-14 14:33                   ` [PATCH v1 01/11] git-gui: allow specifying path '.' to the browser Mark Levedahl
                                     ` (12 more replies)
  0 siblings, 13 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-14 14:33 UTC (permalink / raw)
  To: git; +Cc: egg_mushroomcow, j6t, bootaina702, Mark Levedahl

git gui has a number of inter-related problems that result in problems
during startup from anything but a checked out worktree pointing at a
valid git repository. Some of the symptoms are:
- blame / browser subcommands, and launching gitk, are intended to be
  useful without a worktree, but fail to work.
- unlike git, git-gui is supposed to use the parent directory as a
  worktree if started from the .git subdirectory in the very common
  single worktree + embedded git repository format. This does not
  work.
- git-gui includes a repository picker allowing a user select a worktree
  from a list and/or start a new repo+worktree: this dialog appears at
  unexpected times, masking useful error feedback on configuration
  problems.

This patch series addresses the above issues, substantially rewriting
the blame / browser command line process, the initial repository and
worktree discovery processes, and using git rev-parse when possible to
handle repository / worktree discovery including any specification of
GIT_DIR or GIT_WORK_TREE to reduce the future likelihood of conflict
with command line git. This also allows explicit user control to avoid
the repository picker masking a configuration error.

Note: I question why git-gui ever exports GIT_WORK_TREE. If it is not
empty, that is the current directory when startup is complete and any
git command will use the current directory as the worktree. If empty,
there is no worktree and the current directory should be (and after this
series, is) at the toplevel of the gitdir: again, there is nothing to
communicate to another process. If a process being launched needs a
different worktree, that should be the startup directory given to the
process without changing git-gui's current directory.

Mark Levedahl (11):
  git-gui: allow specifying path '.' to the browser
  git-gui: refactor browser / blame argument parsing
  git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE
  git-gui: put choose_repository::pick in a proc
  git-gui: use --absolute-git-dir
  git gui: GIT_DIR / GIT_WORK_TREE make any discovery error fatal
  git-gui: use rev-parse exclusively to find a repository
  git-gui: simplify [is_bare] to report if a worktree is known
  git-gui: support using repository parent dir as a worktree
  git-gui: improve worktree discovery
  git-gui: add gui and pick as explicit subcommands

 git-gui.sh | 276 ++++++++++++++++++++++++++---------------------------
 1 file changed, 135 insertions(+), 141 deletions(-)

-- 
2.54.0.99.14

^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v1 01/11] git-gui: allow specifying path '.' to the browser
  2026-05-14 14:33                 ` [PATCH v1 00/11] Improve git gui operation without a worktree Mark Levedahl
@ 2026-05-14 14:33                   ` Mark Levedahl
  2026-05-15 15:54                     ` Johannes Sixt
  2026-05-14 14:33                   ` [PATCH v1 02/11] git-gui: refactor browser / blame argument parsing Mark Levedahl
                                     ` (11 subsequent siblings)
  12 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-14 14:33 UTC (permalink / raw)
  To: git; +Cc: egg_mushroomcow, j6t, bootaina702, Mark Levedahl

Invoking "git-gui browser rev ." should show the file browser for the
commitish rev, starting at the root directory. This errors out in
normalize_relpath because the '.' is removed, yielding an empty list as
argument to [file join ...]. Fix this.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/git-gui.sh b/git-gui.sh
index 23fe76e..6048f92 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -2965,7 +2965,11 @@ proc normalize_relpath {path} {
 		}
 		lappend elements $item
 	}
-	return [eval file join $elements]
+	if {$elements ne {}} {
+		return [eval file join $elements]
+	} else {
+		return {}
+	}
 }
 
 # -- Not a normal commit type invocation?  Do that instead!
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 01/11] git-gui: allow specifying path '.' to the browser
  2026-05-14 14:33                   ` [PATCH v1 01/11] git-gui: allow specifying path '.' to the browser Mark Levedahl
@ 2026-05-15 15:54                     ` Johannes Sixt
  2026-05-16 13:38                       ` Mark Levedahl
  0 siblings, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-15 15:54 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 14.05.26 um 16:33 schrieb Mark Levedahl:
> Invoking "git-gui browser rev ." should show the file browser for the
> commitish rev, starting at the root directory. This errors out in
> normalize_relpath because the '.' is removed, yielding an empty list as
> argument to [file join ...]. Fix this.

Good catch!

The description isn't precise, though. '.' means to list the current
directory. The mentioned problem happens only if this is also the root
of the working tree.

> 
> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
> ---
>  git-gui.sh | 6 +++++-
>  1 file changed, 5 insertions(+), 1 deletion(-)
> 
> diff --git a/git-gui.sh b/git-gui.sh
> index 23fe76e..6048f92 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -2965,7 +2965,11 @@ proc normalize_relpath {path} {
>  		}
>  		lappend elements $item
>  	}
> -	return [eval file join $elements]
> +	if {$elements ne {}} {
> +		return [eval file join $elements]
> +	} else {
> +		return {}
> +	}
>  }
>  
>  # -- Not a normal commit type invocation?  Do that instead!


-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 01/11] git-gui: allow specifying path '.' to the browser
  2026-05-15 15:54                     ` Johannes Sixt
@ 2026-05-16 13:38                       ` Mark Levedahl
  0 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-16 13:38 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: egg_mushroomcow, bootaina702, git



On 5/15/26 11:54 AM, Johannes Sixt wrote:
> The description isn't precise, though. '.' means to list the current
> directory. The mentioned problem happens only if this is also the root
> of the working tree.
Easy to fix, will do.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v1 02/11] git-gui: refactor browser / blame argument parsing
  2026-05-14 14:33                 ` [PATCH v1 00/11] Improve git gui operation without a worktree Mark Levedahl
  2026-05-14 14:33                   ` [PATCH v1 01/11] git-gui: allow specifying path '.' to the browser Mark Levedahl
@ 2026-05-14 14:33                   ` Mark Levedahl
  2026-05-15 15:56                     ` Johannes Sixt
  2026-05-14 14:33                   ` [PATCH v1 03/11] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE Mark Levedahl
                                     ` (10 subsequent siblings)
  12 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-14 14:33 UTC (permalink / raw)
  To: git; +Cc: egg_mushroomcow, j6t, bootaina702, Mark Levedahl

git-gui has subcommands blame and browser, both of which accept a
pathname, possibly preceded by a commit-ish item to specify a revision.
Also, blame can take a first argument that gives a line number to focus.

The command line parser for the above is more complex than needed, and
cannot work without a worktree as the pathname objects are checked
against the current worktree for existence. This also precludes naming a
directory or file that does not exist on the currently checked out
branch.

So, replace this with a simpler parser that looks at argument number and
number of arguments to know what value to expect. The blame and browser
backends already have error checking with diagnostic information, so
defer most error checking to those. Also, allow a line-number selection
to be given and silently ignored for the browser, further simplifying
this code.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 66 +++++++++++++-----------------------------------------
 1 file changed, 16 insertions(+), 50 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 6048f92..a951fcd 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -2986,51 +2986,34 @@ blame {
 	set head {}
 	set path {}
 	set jump_spec {}
-	set is_path 0
+	set nargs [llength $argv]
+	if {$nargs < 1} {
+		usage
+	}
+	set argn 0
 	foreach a $argv {
-		set p [file join $_prefix $a]
+		set argn [expr {$argn + 1}]
 
-		if {$is_path || [file exists $p]} {
-			if {$path ne {}} usage
-			set path [normalize_relpath $p]
-			break
-		} elseif {$a eq {--}} {
-			if {$path ne {}} {
-				if {$head ne {}} usage
-				set head $path
-				set path {}
+		if {$argn < $nargs} {
+			# revision or line number
+			if {[regexp {^--line=(\d+)$} $a a lnum]} {
+				set jump_spec [list $lnum]
+			} else {
+				set head $a
 			}
-			set is_path 1
-		} elseif {[regexp {^--line=(\d+)$} $a a lnum]} {
-			if {$jump_spec ne {} || $head ne {}} usage
-			set jump_spec [list $lnum]
-		} elseif {$head eq {}} {
-			if {$head ne {}} usage
-			set head $a
-			set is_path 1
-		} else {
-			usage
-		}
-	}
-	unset is_path
-
-	if {$head ne {} && $path eq {}} {
-		if {[string index $head 0] eq {/}} {
-			set path [normalize_relpath $head]
-			set head {}
 		} else {
-			set path [normalize_relpath $_prefix$head]
-			set head {}
+			set path [normalize_relpath $a]
 		}
 	}
 
 	if {$head eq {}} {
 		load_current_branch
+		set head $current_branch
 	} else {
 		if {[regexp [string map "@@ [expr $hashlength - 1]" {^[0-9a-f]{1,@@}$}] $head]} {
 			if {[catch {
-					set head [git rev-parse --verify $head]
-				} err]} {
+				set head [git rev-parse --verify $head]
+			} err]} {
 				if {[tk windowingsystem] eq "win32"} {
 					tk_messageBox -icon error -title [mc Error] -message $err
 				} else {
@@ -3046,26 +3029,9 @@ blame {
 	switch -- $subcommand {
 	browser {
 		if {$jump_spec ne {}} usage
-		if {$head eq {}} {
-			if {$path ne {} && [file isdirectory $path]} {
-				set head $current_branch
-			} else {
-				set head $path
-				set path {}
-			}
-		}
 		browser::new $head $path
 	}
 	blame   {
-		if {$head eq {} && ![file exists $path]} {
-			catch {wm withdraw .}
-			tk_messageBox \
-				-icon error \
-				-type ok \
-				-title [mc "git-gui: fatal error"] \
-				-message [mc "fatal: cannot stat path %s: No such file or directory" $path]
-			exit 1
-		}
 		blame::new $head $path $jump_spec
 	}
 	}
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 02/11] git-gui: refactor browser / blame argument parsing
  2026-05-14 14:33                   ` [PATCH v1 02/11] git-gui: refactor browser / blame argument parsing Mark Levedahl
@ 2026-05-15 15:56                     ` Johannes Sixt
  2026-05-16 14:21                       ` Mark Levedahl
  0 siblings, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-15 15:56 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 14.05.26 um 16:33 schrieb Mark Levedahl:
> git-gui has subcommands blame and browser, both of which accept a
> pathname, possibly preceded by a commit-ish item to specify a revision.
> Also, blame can take a first argument that gives a line number to focus.
> 
> The command line parser for the above is more complex than needed, and
> cannot work without a worktree as the pathname objects are checked
> against the current worktree for existence. This also precludes naming a
> directory or file that does not exist on the currently checked out
> branch.

While the old browser isn't simple, it implements the strategy "revs
before paths, no revs after the first path or '--'" that is applied by
every git command. The rewritten parser is only slightly simpler. It
should not ignore "--". Furthermore, the old parser ignored excessive
trailing arguments, while the new parser ignores excessive leading
arguments. Neither is desirable, and we should report an incorrect
argument list (if we don't, then I prefer the old behavior).

> 
> So, replace this with a simpler parser that looks at argument number and
> number of arguments to know what value to expect. The blame and browser
> backends already have error checking with diagnostic information, so
> defer most error checking to those. Also, allow a line-number selection
> to be given and silently ignored for the browser, further simplifying
> this code.

The line number selection isn't ignored by the browser, but reported as
an incorrect usage before and after this patch.

> 
> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
> ---
>  git-gui.sh | 66 +++++++++++++-----------------------------------------
>  1 file changed, 16 insertions(+), 50 deletions(-)
> 
> diff --git a/git-gui.sh b/git-gui.sh
> index 6048f92..a951fcd 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -2986,51 +2986,34 @@ blame {
>  	set head {}
>  	set path {}
>  	set jump_spec {}
> -	set is_path 0
> +	set nargs [llength $argv]
> +	if {$nargs < 1} {
> +		usage
> +	}
> +	set argn 0
>  	foreach a $argv {
> -		set p [file join $_prefix $a]
> +		set argn [expr {$argn + 1}]
>  
> -		if {$is_path || [file exists $p]} {
> -			if {$path ne {}} usage
> -			set path [normalize_relpath $p]
> -			break
> -		} elseif {$a eq {--}} {
> -			if {$path ne {}} {
> -				if {$head ne {}} usage
> -				set head $path
> -				set path {}
> +		if {$argn < $nargs} {
> +			# revision or line number
> +			if {[regexp {^--line=(\d+)$} $a a lnum]} {
> +				set jump_spec [list $lnum]
> +			} else {
> +				set head $a
>  			}
> -			set is_path 1
> -		} elseif {[regexp {^--line=(\d+)$} $a a lnum]} {
> -			if {$jump_spec ne {} || $head ne {}} usage
> -			set jump_spec [list $lnum]
> -		} elseif {$head eq {}} {
> -			if {$head ne {}} usage
> -			set head $a
> -			set is_path 1
> -		} else {
> -			usage
> -		}
> -	}
> -	unset is_path
> -
> -	if {$head ne {} && $path eq {}} {
> -		if {[string index $head 0] eq {/}} {
> -			set path [normalize_relpath $head]
> -			set head {}
>  		} else {
> -			set path [normalize_relpath $_prefix$head]
> -			set head {}
> +			set path [normalize_relpath $a]

This loses the capability to request a browser relative to a
subdirectory of the working tree. For example, "git gui browser main ."
now shows the top-level directory instead of the subdirectory when
invoked from a subdirectory.

>  		}
>  	}
>  
>  	if {$head eq {}} {
>  		load_current_branch
> +		set head $current_branch
>  	} else {
>  		if {[regexp [string map "@@ [expr $hashlength - 1]" {^[0-9a-f]{1,@@}$}] $head]} {
>  			if {[catch {
> -					set head [git rev-parse --verify $head]
> -				} err]} {
> +				set head [git rev-parse --verify $head]
> +			} err]} {

Please leave the indentation unchanged.

>  				if {[tk windowingsystem] eq "win32"} {
>  					tk_messageBox -icon error -title [mc Error] -message $err
>  				} else {
> @@ -3046,26 +3029,9 @@ blame {
>  	switch -- $subcommand {
>  	browser {
>  		if {$jump_spec ne {}} usage
> -		if {$head eq {}} {
> -			if {$path ne {} && [file isdirectory $path]} {
> -				set head $current_branch
> -			} else {
> -				set head $path
> -				set path {}
> -			}
> -		}
>  		browser::new $head $path
>  	}
>  	blame   {
> -		if {$head eq {} && ![file exists $path]} {
> -			catch {wm withdraw .}
> -			tk_messageBox \
> -				-icon error \
> -				-type ok \
> -				-title [mc "git-gui: fatal error"] \
> -				-message [mc "fatal: cannot stat path %s: No such file or directory" $path]
> -			exit 1
> -		}
>  		blame::new $head $path $jump_spec
>  	}
>  	}

The check for the existence of files is actually necessary to
disambiguate the meaning of the argument. If a file "maint" exists, then
the argument is to be interpreted as path, not as the ref "maint", even
if that exists, too.

I suggest to protect the "file exists" calls with ($_gitworktree ne {}
&& ...) or (![is_bare] && ...) to handle being invoked from a bare
repository. That is, in a bare repository we treat arguments the same as
files that do not exist in the currently checked-out branch.

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 02/11] git-gui: refactor browser / blame argument parsing
  2026-05-15 15:56                     ` Johannes Sixt
@ 2026-05-16 14:21                       ` Mark Levedahl
  0 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-16 14:21 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: egg_mushroomcow, bootaina702, git



On 5/15/26 11:56 AM, Johannes Sixt wrote:
snip...
> The check for the existence of files is actually necessary to
> disambiguate the meaning of the argument. If a file "maint" exists, then
> the argument is to be interpreted as path, not as the ref "maint", even
> if that exists, too.
>
> I suggest to protect the "file exists" calls with ($_gitworktree ne {}
> && ...) or (![is_bare] && ...) to handle being invoked from a bare
> repository. That is, in a bare repository we treat arguments the same as
> files that do not exist in the currently checked-out branch.

Let me start over, addressing only the use-case in a bare repository.

Mark


^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v1 03/11] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE
  2026-05-14 14:33                 ` [PATCH v1 00/11] Improve git gui operation without a worktree Mark Levedahl
  2026-05-14 14:33                   ` [PATCH v1 01/11] git-gui: allow specifying path '.' to the browser Mark Levedahl
  2026-05-14 14:33                   ` [PATCH v1 02/11] git-gui: refactor browser / blame argument parsing Mark Levedahl
@ 2026-05-14 14:33                   ` Mark Levedahl
  2026-05-15 15:58                     ` Johannes Sixt
  2026-05-14 14:33                   ` [PATCH v1 04/11] git-gui: put choose_repository::pick in a proc Mark Levedahl
                                     ` (9 subsequent siblings)
  12 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-14 14:33 UTC (permalink / raw)
  To: git; +Cc: egg_mushroomcow, j6t, bootaina702, Mark Levedahl

git-gui unconditionally exports GIT_DIR and GIT_WORK_TREE to the
environment, and furthmore unconditionally unsets these in many places.
But, GIT_WORK_TREE should be set only if it is not {} as the empty
value, really meaning no work-tree is found, causes git to throw fatal
errors (git-gui gets the error from branch --show-current).  Fixing this
is required to allow blame and browser to operate from a repository
without a worktree.

Establish a pair of functions to remove GIT_DIR and GIT_WORK_TREE from
the environment, avoiding any error if they do not exist. Also, add a
function to export these, but export GIT_WORK_TREE only if not empty.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 32 ++++++++++++++++++++++----------
 1 file changed, 22 insertions(+), 10 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index a951fcd..387cad6 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1122,6 +1122,22 @@ unset argv0dir
 ##
 ## repository setup
 
+proc set_gitdir_vars {} {
+	global _gitdir _gitworktree env
+	if {$_gitdir ne {}} {
+		set env(GIT_DIR) $_gitdir
+	}
+	if {$_gitworktree ne {}} {
+		set env(GIT_WORK_TREE) $_gitworktree
+	}
+}
+
+proc unset_gitdir_vars {} {
+	global env
+	catch {unset env(GIT_DIR)}
+	catch {unset env(GIT_WORK_TREE)}
+}
+
 set picked 0
 if {[catch {
 		set _gitdir $env(GIT_DIR)
@@ -1207,8 +1223,8 @@ if {[lindex $_reponame end] eq {.git}} {
 	set _reponame [lindex $_reponame end]
 }
 
-set env(GIT_DIR) $_gitdir
-set env(GIT_WORK_TREE) $_gitworktree
+# Export the final paths
+set_gitdir_vars
 
 ######################################################################
 ##
@@ -2050,13 +2066,11 @@ proc do_gitk {revs {is_submodule false}} {
 			# TODO we could make life easier (start up faster?) for gitk
 			# by setting these to the appropriate values to allow gitk
 			# to skip the heuristics to find their proper value
-			unset env(GIT_DIR)
-			unset env(GIT_WORK_TREE)
+			unset_gitdir_vars
 		}
 		safe_exec_bg [concat $cmd $revs "--" "--"]
 
-		set env(GIT_DIR) $_gitdir
-		set env(GIT_WORK_TREE) $_gitworktree
+		set_gitdir_vars
 		cd $pwd
 
 		if {[info exists main_status]} {
@@ -2084,16 +2098,14 @@ proc do_git_gui {} {
 
 		# see note in do_gitk about unsetting these vars when
 		# running tools in a submodule
-		unset env(GIT_DIR)
-		unset env(GIT_WORK_TREE)
+		unset_gitdir_vars
 
 		set pwd [pwd]
 		cd $current_diff_path
 
 		safe_exec_bg [concat $exe gui]
 
-		set env(GIT_DIR) $_gitdir
-		set env(GIT_WORK_TREE) $_gitworktree
+		set_gitdir_vars
 		cd $pwd
 
 		set status_operation [$::main_status \
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 03/11] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE
  2026-05-14 14:33                   ` [PATCH v1 03/11] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE Mark Levedahl
@ 2026-05-15 15:58                     ` Johannes Sixt
  2026-05-16 14:25                       ` Mark Levedahl
  0 siblings, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-15 15:58 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 14.05.26 um 16:33 schrieb Mark Levedahl:
> git-gui unconditionally exports GIT_DIR and GIT_WORK_TREE to the
> environment, and furthmore unconditionally unsets these in many places.
> But, GIT_WORK_TREE should be set only if it is not {} as the empty
> value, really meaning no work-tree is found, causes git to throw fatal
> errors (git-gui gets the error from branch --show-current).  Fixing this
> is required to allow blame and browser to operate from a repository
> without a worktree.
> 
> Establish a pair of functions to remove GIT_DIR and GIT_WORK_TREE from
> the environment, avoiding any error if they do not exist. Also, add a
> function to export these, but export GIT_WORK_TREE only if not empty.

Good. But as I said in a parallel thread, I actually concur with your
assessment in the coverletter of this patch series that GIT_WORK_TREE
should be not set at all. At least in the modes that require a working tree.

> 
> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
> ---
>  git-gui.sh | 32 ++++++++++++++++++++++----------
>  1 file changed, 22 insertions(+), 10 deletions(-)
> 
> diff --git a/git-gui.sh b/git-gui.sh
> index a951fcd..387cad6 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -1122,6 +1122,22 @@ unset argv0dir
>  ##
>  ## repository setup
>  
> +proc set_gitdir_vars {} {
> +	global _gitdir _gitworktree env
> +	if {$_gitdir ne {}} {
> +		set env(GIT_DIR) $_gitdir
> +	}
> +	if {$_gitworktree ne {}} {
> +		set env(GIT_WORK_TREE) $_gitworktree
> +	}
> +}
> +
> +proc unset_gitdir_vars {} {
> +	global env
> +	catch {unset env(GIT_DIR)}
> +	catch {unset env(GIT_WORK_TREE)}
> +}
> +
>  set picked 0
>  if {[catch {
>  		set _gitdir $env(GIT_DIR)
> @@ -1207,8 +1223,8 @@ if {[lindex $_reponame end] eq {.git}} {
>  	set _reponame [lindex $_reponame end]
>  }
>  
> -set env(GIT_DIR) $_gitdir
> -set env(GIT_WORK_TREE) $_gitworktree
> +# Export the final paths
> +set_gitdir_vars
>  
>  ######################################################################
>  ##
> @@ -2050,13 +2066,11 @@ proc do_gitk {revs {is_submodule false}} {
>  			# TODO we could make life easier (start up faster?) for gitk
>  			# by setting these to the appropriate values to allow gitk
>  			# to skip the heuristics to find their proper value
> -			unset env(GIT_DIR)
> -			unset env(GIT_WORK_TREE)
> +			unset_gitdir_vars
>  		}
>  		safe_exec_bg [concat $cmd $revs "--" "--"]
>  
> -		set env(GIT_DIR) $_gitdir
> -		set env(GIT_WORK_TREE) $_gitworktree
> +		set_gitdir_vars
>  		cd $pwd
>  
>  		if {[info exists main_status]} {
> @@ -2084,16 +2098,14 @@ proc do_git_gui {} {
>  
>  		# see note in do_gitk about unsetting these vars when
>  		# running tools in a submodule
> -		unset env(GIT_DIR)
> -		unset env(GIT_WORK_TREE)
> +		unset_gitdir_vars
>  
>  		set pwd [pwd]
>  		cd $current_diff_path
>  
>  		safe_exec_bg [concat $exe gui]
>  
> -		set env(GIT_DIR) $_gitdir
> -		set env(GIT_WORK_TREE) $_gitworktree
> +		set_gitdir_vars
>  		cd $pwd
>  
>  		set status_operation [$::main_status \

After these changes, a 'global env' probably becomes stale and could be
removed.

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 03/11] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE
  2026-05-15 15:58                     ` Johannes Sixt
@ 2026-05-16 14:25                       ` Mark Levedahl
  0 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-16 14:25 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: egg_mushroomcow, bootaina702, git



On 5/15/26 11:58 AM, Johannes Sixt wrote:
> Am 14.05.26 um 16:33 schrieb Mark Levedahl:
>> git-gui unconditionally exports GIT_DIR and GIT_WORK_TREE to the
>> environment, and furthmore unconditionally unsets these in many places.
>> But, GIT_WORK_TREE should be set only if it is not {} as the empty
>> value, really meaning no work-tree is found, causes git to throw fatal
>> errors (git-gui gets the error from branch --show-current).  Fixing this
>> is required to allow blame and browser to operate from a repository
>> without a worktree.
>>
>> Establish a pair of functions to remove GIT_DIR and GIT_WORK_TREE from
>> the environment, avoiding any error if they do not exist. Also, add a
>> function to export these, but export GIT_WORK_TREE only if not empty.
> Good. But as I said in a parallel thread, I actually concur with your
> assessment in the coverletter of this patch series that GIT_WORK_TREE
> should be not set at all. At least in the modes that require a working tree.
>
>> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
>> ---
>>  git-gui.sh | 32 ++++++++++++++++++++++----------
>>  1 file changed, 22 insertions(+), 10 deletions(-)
>>
>> diff --git a/git-gui.sh b/git-gui.sh
>> index a951fcd..387cad6 100755
>> --- a/git-gui.sh
>> +++ b/git-gui.sh
>> @@ -1122,6 +1122,22 @@ unset argv0dir
>>  ##
>>  ## repository setup
>>  
>> +proc set_gitdir_vars {} {
>> +	global _gitdir _gitworktree env
>> +	if {$_gitdir ne {}} {
>> +		set env(GIT_DIR) $_gitdir
>> +	}
>> +	if {$_gitworktree ne {}} {
>> +		set env(GIT_WORK_TREE) $_gitworktree
>> +	}
>> +}
>> +
>> +proc unset_gitdir_vars {} {
>> +	global env
>> +	catch {unset env(GIT_DIR)}
>> +	catch {unset env(GIT_WORK_TREE)}
>> +}
>> +
>>  set picked 0
>>  if {[catch {
>>  		set _gitdir $env(GIT_DIR)
>> @@ -1207,8 +1223,8 @@ if {[lindex $_reponame end] eq {.git}} {
>>  	set _reponame [lindex $_reponame end]
>>  }
>>  
>> -set env(GIT_DIR) $_gitdir
>> -set env(GIT_WORK_TREE) $_gitworktree
>> +# Export the final paths
>> +set_gitdir_vars
>>  
>>  ######################################################################
>>  ##
>> @@ -2050,13 +2066,11 @@ proc do_gitk {revs {is_submodule false}} {
>>  			# TODO we could make life easier (start up faster?) for gitk
>>  			# by setting these to the appropriate values to allow gitk
>>  			# to skip the heuristics to find their proper value
>> -			unset env(GIT_DIR)
>> -			unset env(GIT_WORK_TREE)
>> +			unset_gitdir_vars
>>  		}
>>  		safe_exec_bg [concat $cmd $revs "--" "--"]
>>  
>> -		set env(GIT_DIR) $_gitdir
>> -		set env(GIT_WORK_TREE) $_gitworktree
>> +		set_gitdir_vars
>>  		cd $pwd
>>  
>>  		if {[info exists main_status]} {
>> @@ -2084,16 +2098,14 @@ proc do_git_gui {} {
>>  
>>  		# see note in do_gitk about unsetting these vars when
>>  		# running tools in a submodule
>> -		unset env(GIT_DIR)
>> -		unset env(GIT_WORK_TREE)
>> +		unset_gitdir_vars
>>  
>>  		set pwd [pwd]
>>  		cd $current_diff_path
>>  
>>  		safe_exec_bg [concat $exe gui]
>>  
>> -		set env(GIT_DIR) $_gitdir
>> -		set env(GIT_WORK_TREE) $_gitworktree
>> +		set_gitdir_vars
>>  		cd $pwd
>>  
>>  		set status_operation [$::main_status \
> After these changes, a 'global env' probably becomes stale and could be
> removed.
>
> -- Hannes
>
Will update to only ever set GIT_DIR, will still remove GIT_WORK_TREE on unset.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v1 04/11] git-gui: put choose_repository::pick in a proc
  2026-05-14 14:33                 ` [PATCH v1 00/11] Improve git gui operation without a worktree Mark Levedahl
                                     ` (2 preceding siblings ...)
  2026-05-14 14:33                   ` [PATCH v1 03/11] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE Mark Levedahl
@ 2026-05-14 14:33                   ` Mark Levedahl
  2026-05-15 11:00                     ` Aina Boot
  2026-05-15 15:59                     ` Johannes Sixt
  2026-05-14 14:33                   ` [PATCH v1 05/11] git-gui: use --absolute-git-dir Mark Levedahl
                                     ` (8 subsequent siblings)
  12 siblings, 2 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-14 14:33 UTC (permalink / raw)
  To: git; +Cc: egg_mushroomcow, j6t, bootaina702, Mark Levedahl

git-gui includes a 'repository picker', which allows creating a new
repository + worktree, or selecting a worktree from a recent list.
git-gui runs the picker when a valid git repository is not found. All of
the code for this is embedded in the discovery process block, making the
latter more difficult to read, and also making things more difficult if
we want to have an explicit 'pick' subcommand to force this to run.

Let's move this invocation and supporting code to a separate proc,
aiding in subsequent refactoring. Assure GIT_DIR and GIT_WORK_TREE are
unset, configuration is loaded, ant that _gitdir is correctly set
afterwards. As this is invoked before worktree discovery, later code
will set that anyway so need not be included here.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 18 +++++++++++-------
 1 file changed, 11 insertions(+), 7 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 387cad6..0b73c35 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1139,6 +1139,16 @@ proc unset_gitdir_vars {} {
 }
 
 set picked 0
+proc pick_repo {} {
+	unset_gitdir_vars
+	load_config 1
+	apply_config
+	choose_repository::pick
+	set _gitdir [git rev-parse --absolute-git-dir]
+	set _prefix {}
+	set picked 1
+}
+
 if {[catch {
 		set _gitdir $env(GIT_DIR)
 		set _prefix {}
@@ -1149,13 +1159,7 @@ if {[catch {
 		set _gitdir [git rev-parse --git-dir]
 		set _prefix [git rev-parse --show-prefix]
 	} err]} {
-	load_config 1
-	apply_config
-	choose_repository::pick
-	if {![file isdirectory $_gitdir]} {
-		exit 1
-	}
-	set picked 1
+		pick_repo
 }
 
 # Use object format as hash algorithm (either "sha1" or "sha256")
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 04/11] git-gui: put choose_repository::pick in a proc
  2026-05-14 14:33                   ` [PATCH v1 04/11] git-gui: put choose_repository::pick in a proc Mark Levedahl
@ 2026-05-15 11:00                     ` Aina Boot
  2026-05-15 13:33                       ` Mark Levedahl
  2026-05-15 15:59                     ` Johannes Sixt
  1 sibling, 1 reply; 134+ messages in thread
From: Aina Boot @ 2026-05-15 11:00 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: Johannes Sixt, Shroom Moo, git

On 5/14/26 2:33 PM, Mark Levedahl wrote:
>  set picked 0
> +proc pick_repo {} {
> +	unset_gitdir_vars
> +	load_config 1
> +	apply_config
> +	choose_repository::pick
> +	set _gitdir [git rev-parse --absolute-git-dir]
> +	set _prefix {}
> +	set picked 1
> +}
> +
>

Here inside the proc it create vars locally, "global..." is missing.

Aina

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 04/11] git-gui: put choose_repository::pick in a proc
  2026-05-15 11:00                     ` Aina Boot
@ 2026-05-15 13:33                       ` Mark Levedahl
  0 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-15 13:33 UTC (permalink / raw)
  To: Aina Boot; +Cc: Johannes Sixt, Shroom Moo, git



On 5/15/26 7:00 AM, Aina Boot wrote:
> On 5/14/26 2:33 PM, Mark Levedahl wrote:
>>  set picked 0
>> +proc pick_repo {} {
>> +	unset_gitdir_vars
>> +	load_config 1
>> +	apply_config
>> +	choose_repository::pick
>> +	set _gitdir [git rev-parse --absolute-git-dir]
>> +	set _prefix {}
>> +	set picked 1
>> +}
>> +
>>
> Here inside the proc it create vars locally, "global..." is missing.
>
> Aina
Yes, also, I missed copying the check on the return variable ::_gitdir. That proc should be:

proc pick_repo {} {
    global _gitdir picked
    unset_gitdir_vars
    load_config 1
    apply_config
    choose_repository::pick
    if {![file isdirectory $_gitdir]} {
        exit 1
    }
    set _gitdir [git rev-parse --absolute-git-dir]
    set picked 1
}

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 04/11] git-gui: put choose_repository::pick in a proc
  2026-05-14 14:33                   ` [PATCH v1 04/11] git-gui: put choose_repository::pick in a proc Mark Levedahl
  2026-05-15 11:00                     ` Aina Boot
@ 2026-05-15 15:59                     ` Johannes Sixt
  2026-05-16 14:29                       ` Mark Levedahl
  1 sibling, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-15 15:59 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 14.05.26 um 16:33 schrieb Mark Levedahl:
> git-gui includes a 'repository picker', which allows creating a new
> repository + worktree, or selecting a worktree from a recent list.
> git-gui runs the picker when a valid git repository is not found. All of
> the code for this is embedded in the discovery process block, making the
> latter more difficult to read, and also making things more difficult if
> we want to have an explicit 'pick' subcommand to force this to run.

OK, let's see how useful this becomes.

> 
> Let's move this invocation and supporting code to a separate proc,
> aiding in subsequent refactoring. Assure GIT_DIR and GIT_WORK_TREE are
> unset, configuration is loaded, ant that _gitdir is correctly set

s/ant/and/

> afterwards. As this is invoked before worktree discovery, later code
> will set that anyway so need not be included here.
> 
> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
> ---
>  git-gui.sh | 18 +++++++++++-------
>  1 file changed, 11 insertions(+), 7 deletions(-)
> 
> diff --git a/git-gui.sh b/git-gui.sh
> index 387cad6..0b73c35 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -1139,6 +1139,16 @@ proc unset_gitdir_vars {} {
>  }
>  
>  set picked 0
> +proc pick_repo {} {
> +	unset_gitdir_vars
> +	load_config 1
> +	apply_config
> +	choose_repository::pick
> +	set _gitdir [git rev-parse --absolute-git-dir]
> +	set _prefix {}
> +	set picked 1
> +}
> +

So, this isn't intended as a plain move of code? Since we set _gitdir
here, we could remove the corresonding lines from lib/choose_repository.tcl.

Is the variable "picked" only needed for this particular picker
invocation? Then it should not be set in the function, but at the call site.

>  if {[catch {
>  		set _gitdir $env(GIT_DIR)
>  		set _prefix {}
> @@ -1149,13 +1159,7 @@ if {[catch {
>  		set _gitdir [git rev-parse --git-dir]
>  		set _prefix [git rev-parse --show-prefix]
>  	} err]} {
> -	load_config 1
> -	apply_config
> -	choose_repository::pick
> -	if {![file isdirectory $_gitdir]} {
> -		exit 1
> -	}
> -	set picked 1
> +		pick_repo

The indentation is off here.

>  }
>  
>  # Use object format as hash algorithm (either "sha1" or "sha256")

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 04/11] git-gui: put choose_repository::pick in a proc
  2026-05-15 15:59                     ` Johannes Sixt
@ 2026-05-16 14:29                       ` Mark Levedahl
  0 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-16 14:29 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: egg_mushroomcow, bootaina702, git



On 5/15/26 11:59 AM, Johannes Sixt wrote:
> Am 14.05.26 um 16:33 schrieb Mark Levedahl:
>> git-gui includes a 'repository picker', which allows creating a new
>> repository + worktree, or selecting a worktree from a recent list.
>> git-gui runs the picker when a valid git repository is not found. All of
>> the code for this is embedded in the discovery process block, making the
>> latter more difficult to read, and also making things more difficult if
>> we want to have an explicit 'pick' subcommand to force this to run.
> OK, let's see how useful this becomes.
>
>> Let's move this invocation and supporting code to a separate proc,
>> aiding in subsequent refactoring. Assure GIT_DIR and GIT_WORK_TREE are
>> unset, configuration is loaded, ant that _gitdir is correctly set
> s/ant/and/
will fix
>> afterwards. As this is invoked before worktree discovery, later code
>> will set that anyway so need not be included here.
>>
>> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
>> ---
>>  git-gui.sh | 18 +++++++++++-------
>>  1 file changed, 11 insertions(+), 7 deletions(-)
>>
>> diff --git a/git-gui.sh b/git-gui.sh
>> index 387cad6..0b73c35 100755
>> --- a/git-gui.sh
>> +++ b/git-gui.sh
>> @@ -1139,6 +1139,16 @@ proc unset_gitdir_vars {} {
>>  }
>>  
>>  set picked 0
>> +proc pick_repo {} {
>> +	unset_gitdir_vars
>> +	load_config 1
>> +	apply_config
>> +	choose_repository::pick
>> +	set _gitdir [git rev-parse --absolute-git-dir]
>> +	set _prefix {}
>> +	set picked 1
>> +}
>> +
> So, this isn't intended as a plain move of code? Since we set _gitdir
> here, we could remove the corresonding lines from lib/choose_repository.tcl.
it should be a plain move.
> Is the variable "picked" only needed for this particular picker
> invocation? Then it should not be set in the function, but at the call site.
I need to better understand how "picked" is used to decide... will do before an update.
>>  if {[catch {
>>  		set _gitdir $env(GIT_DIR)
>>  		set _prefix {}
>> @@ -1149,13 +1159,7 @@ if {[catch {
>>  		set _gitdir [git rev-parse --git-dir]
>>  		set _prefix [git rev-parse --show-prefix]
>>  	} err]} {
>> -	load_config 1
>> -	apply_config
>> -	choose_repository::pick
>> -	if {![file isdirectory $_gitdir]} {
>> -		exit 1
>> -	}
>> -	set picked 1
>> +		pick_repo
> The indentation is off here.
will fix.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v1 05/11] git-gui: use --absolute-git-dir
  2026-05-14 14:33                 ` [PATCH v1 00/11] Improve git gui operation without a worktree Mark Levedahl
                                     ` (3 preceding siblings ...)
  2026-05-14 14:33                   ` [PATCH v1 04/11] git-gui: put choose_repository::pick in a proc Mark Levedahl
@ 2026-05-14 14:33                   ` Mark Levedahl
  2026-05-15 16:00                     ` Johannes Sixt
  2026-05-14 14:33                   ` [PATCH v1 06/11] git gui: GIT_DIR / GIT_WORK_TREE make any discovery error fatal Mark Levedahl
                                     ` (7 subsequent siblings)
  12 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-14 14:33 UTC (permalink / raw)
  To: git; +Cc: egg_mushroomcow, j6t, bootaina702, Mark Levedahl

git-gui uses git rev-parse --git-dir to get the pathname of the
discovered git repository. The returned value can be relative, and is
'.' if the current directory is the top of the repository directory
itself.  git-gui has code to change '.' to [pwd] in this case so that
subsequent logic runs.

But, git rev-parse supports --absolute-git-dir from fac60b8925
("rev-parse: add option for absolute or relative path formatting",
2020-12-13), and included in git 2.31. git-gui requires git >= 2.36, so
this more useful form is always available. Use --absolute-git-dir to
always get an absolute path, avoiding the need for other checks.
---
 git-gui.sh | 10 ++--------
 1 file changed, 2 insertions(+), 8 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 0b73c35..c2cf5f1 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1156,7 +1156,7 @@ if {[catch {
 	&& [catch {
 		# beware that from the .git dir this sets _gitdir to .
 		# and _prefix to the empty string
-		set _gitdir [git rev-parse --git-dir]
+		set _gitdir [git rev-parse --absolute-git-dir]
 		set _prefix [git rev-parse --show-prefix]
 	} err]} {
 		pick_repo
@@ -1173,18 +1173,12 @@ if {$hashalgorithm eq "sha1"} {
 	exit 1
 }
 
-# we expand the _gitdir when it's just a single dot (i.e. when we're being
-# run from the .git dir itself) lest the routines to find the worktree
-# get confused
-if {$_gitdir eq "."} {
-	set _gitdir [pwd]
-}
-
 if {![file isdirectory $_gitdir]} {
 	catch {wm withdraw .}
 	error_popup [strcat [mc "Git directory not found:"] "\n\n$_gitdir"]
 	exit 1
 }
+
 # _gitdir exists, so try loading the config
 load_config 0
 apply_config
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 05/11] git-gui: use --absolute-git-dir
  2026-05-14 14:33                   ` [PATCH v1 05/11] git-gui: use --absolute-git-dir Mark Levedahl
@ 2026-05-15 16:00                     ` Johannes Sixt
  2026-05-16 14:33                       ` Mark Levedahl
  0 siblings, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-15 16:00 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 14.05.26 um 16:33 schrieb Mark Levedahl:
> git-gui uses git rev-parse --git-dir to get the pathname of the
> discovered git repository. The returned value can be relative, and is
> '.' if the current directory is the top of the repository directory
> itself.  git-gui has code to change '.' to [pwd] in this case so that
> subsequent logic runs.
> 
> But, git rev-parse supports --absolute-git-dir from fac60b8925
> ("rev-parse: add option for absolute or relative path formatting",
> 2020-12-13), and included in git 2.31. git-gui requires git >= 2.36, so
> this more useful form is always available. Use --absolute-git-dir to
> always get an absolute path, avoiding the need for other checks.

Nice!

However, the patch is incomplete. We set _gitdir also from
lib/choose_repository.tcl. I think it would be best to swap this patch
with patch 4/11, remove the _gitdir setters from the picker
implementation, and call `rev-parse --absolute-git-dir` like you did in
4/11. This depends on that the picker sets the current directory to the
top-level of the working tree with the embeded .git directory.

BTW, missing sign-off.

> ---
>  git-gui.sh | 10 ++--------
>  1 file changed, 2 insertions(+), 8 deletions(-)
> 
> diff --git a/git-gui.sh b/git-gui.sh
> index 0b73c35..c2cf5f1 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -1156,7 +1156,7 @@ if {[catch {
>  	&& [catch {
>  		# beware that from the .git dir this sets _gitdir to .
>  		# and _prefix to the empty string
> -		set _gitdir [git rev-parse --git-dir]
> +		set _gitdir [git rev-parse --absolute-git-dir]
>  		set _prefix [git rev-parse --show-prefix]
>  	} err]} {
>  		pick_repo
> @@ -1173,18 +1173,12 @@ if {$hashalgorithm eq "sha1"} {
>  	exit 1
>  }
>  
> -# we expand the _gitdir when it's just a single dot (i.e. when we're being
> -# run from the .git dir itself) lest the routines to find the worktree
> -# get confused
> -if {$_gitdir eq "."} {
> -	set _gitdir [pwd]
> -}
> -
>  if {![file isdirectory $_gitdir]} {
>  	catch {wm withdraw .}
>  	error_popup [strcat [mc "Git directory not found:"] "\n\n$_gitdir"]
>  	exit 1
>  }
> +
>  # _gitdir exists, so try loading the config
>  load_config 0
>  apply_config

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 05/11] git-gui: use --absolute-git-dir
  2026-05-15 16:00                     ` Johannes Sixt
@ 2026-05-16 14:33                       ` Mark Levedahl
  0 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-16 14:33 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: egg_mushroomcow, bootaina702, git



On 5/15/26 12:00 PM, Johannes Sixt wrote:
> Am 14.05.26 um 16:33 schrieb Mark Levedahl:
>> git-gui uses git rev-parse --git-dir to get the pathname of the
>> discovered git repository. The returned value can be relative, and is
>> '.' if the current directory is the top of the repository directory
>> itself.  git-gui has code to change '.' to [pwd] in this case so that
>> subsequent logic runs.
>>
>> But, git rev-parse supports --absolute-git-dir from fac60b8925
>> ("rev-parse: add option for absolute or relative path formatting",
>> 2020-12-13), and included in git 2.31. git-gui requires git >= 2.36, so
>> this more useful form is always available. Use --absolute-git-dir to
>> always get an absolute path, avoiding the need for other checks.
> Nice!
>
> However, the patch is incomplete. We set _gitdir also from
> lib/choose_repository.tcl. I think it would be best to swap this patch
> with patch 4/11, remove the _gitdir setters from the picker
> implementation, and call `rev-parse --absolute-git-dir` like you did in
> 4/11. This depends on that the picker sets the current directory to the
> top-level of the working tree with the embeded .git directory.
>
> BTW, missing sign-off.
I will change the interface to the picker so that success / failure is a returned value
rather than _gitdir being non-empty, then rework order and content of these patches.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v1 06/11] git gui: GIT_DIR / GIT_WORK_TREE make any discovery error fatal
  2026-05-14 14:33                 ` [PATCH v1 00/11] Improve git gui operation without a worktree Mark Levedahl
                                     ` (4 preceding siblings ...)
  2026-05-14 14:33                   ` [PATCH v1 05/11] git-gui: use --absolute-git-dir Mark Levedahl
@ 2026-05-14 14:33                   ` Mark Levedahl
  2026-05-14 14:33                   ` [PATCH v1 07/11] git-gui: use rev-parse exclusively to find a repository Mark Levedahl
                                     ` (6 subsequent siblings)
  12 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-14 14:33 UTC (permalink / raw)
  To: git; +Cc: egg_mushroomcow, j6t, bootaina702, Mark Levedahl

git accepts any combination of GIT_DIR and GIT_WORK_TREE to override the
normal repository and worktree discovery process.  git-gui should accept
any such valid configuration, but overriding the discovery process means
the user has assured that the combination of current directory, GIT_DIR,
and GIT_WORK_TREE will lead to the correct repository and worktree. As
such, an error found during discovery where either or both of GIT_DIR
and GIT_WORK_TREE are set is a fatal error, no further exploration
should be tried.

Provide a common proc to support displaying an error message and exiting
if GIT_DIR or GIT_WORK_TREE are in the environment.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/git-gui.sh b/git-gui.sh
index c2cf5f1..2e2ddc0 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1122,6 +1122,24 @@ unset argv0dir
 ##
 ## repository setup
 
+proc is_gitvars_error {err} {
+	set havevars 0
+	set GIT_DIR {}
+	set GIT_WORK_TREE {}
+	catch {set GIT_DIR $::env(GIT_DIR); set havevars 1}
+	catch {set GIT_WORK_TREE $::env(GIT_WORK_TREE) ; set havevars 1}
+
+	if {$havevars} {
+		catch {wm withdraw .}
+		error_popup [strcat [mc "Invalid configuration:"] \
+		   "\n" "GIT_DIR: " $GIT_DIR \
+		   "\n" "GIT_WORK_TREE: " $GIT_WORK_TREE \
+			"\n\n$err"]
+		return 1
+	}
+	return 0
+}
+
 proc set_gitdir_vars {} {
 	global _gitdir _gitworktree env
 	if {$_gitdir ne {}} {
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* [PATCH v1 07/11] git-gui: use rev-parse exclusively to find a repository
  2026-05-14 14:33                 ` [PATCH v1 00/11] Improve git gui operation without a worktree Mark Levedahl
                                     ` (5 preceding siblings ...)
  2026-05-14 14:33                   ` [PATCH v1 06/11] git gui: GIT_DIR / GIT_WORK_TREE make any discovery error fatal Mark Levedahl
@ 2026-05-14 14:33                   ` Mark Levedahl
  2026-05-15 16:06                     ` Johannes Sixt
  2026-05-14 14:33                   ` [PATCH v1 08/11] git-gui: simplify [is_bare] to report if a worktree is known Mark Levedahl
                                     ` (5 subsequent siblings)
  12 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-14 14:33 UTC (permalink / raw)
  To: git; +Cc: egg_mushroomcow, j6t, bootaina702, Mark Levedahl

git-gui attempts to use env(GIT_DIR) directly as the git repository,
accepting GIT_DIR if it is a directory. Only if that fails is git
rev-parse used to discover the repository.  But, this avoids all of
git-core's validity checking on a repository, thus possibly deferring an
error to a later step, possibly unexpected. Repository validation should
be part of initial setup so that later processing does not need error
trapping for configuration errors.

Let's just invoke rev-parse so all error checking is done. Stop here if
the user set GIT_DIR or GIT_WORK_TREE. Otherwise, continue the existing
behavior and show the repository picker.

Also, remove a later check on whether _gitdir is a directory: that code
cannot be reached without rev-parse having validating the repository.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 24 +++++++++---------------
 1 file changed, 9 insertions(+), 15 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 2e2ddc0..81789dd 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -374,6 +374,7 @@ set _gitdir {}
 set _gitworktree {}
 set _isbare {}
 set _githtmldir {}
+set _prefix {}
 set _reponame {}
 set _shellpath {@@SHELL_PATH@@}
 
@@ -1167,19 +1168,18 @@ proc pick_repo {} {
 	set picked 1
 }
 
+# find repository.
 if {[catch {
-		set _gitdir $env(GIT_DIR)
-		set _prefix {}
-		}]
-	&& [catch {
-		# beware that from the .git dir this sets _gitdir to .
-		# and _prefix to the empty string
-		set _gitdir [git rev-parse --absolute-git-dir]
-		set _prefix [git rev-parse --show-prefix]
-	} err]} {
+	set _gitdir [git rev-parse --absolute-git-dir]
+} err]} {
+	if {[is_gitvars_error $err]} {
+		exit 1
+	} else {
 		pick_repo
+	}
 }
 
+
 # Use object format as hash algorithm (either "sha1" or "sha256")
 set hashalgorithm [git rev-parse --show-object-format]
 if {$hashalgorithm eq "sha1"} {
@@ -1191,12 +1191,6 @@ if {$hashalgorithm eq "sha1"} {
 	exit 1
 }
 
-if {![file isdirectory $_gitdir]} {
-	catch {wm withdraw .}
-	error_popup [strcat [mc "Git directory not found:"] "\n\n$_gitdir"]
-	exit 1
-}
-
 # _gitdir exists, so try loading the config
 load_config 0
 apply_config
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 07/11] git-gui: use rev-parse exclusively to find a repository
  2026-05-14 14:33                   ` [PATCH v1 07/11] git-gui: use rev-parse exclusively to find a repository Mark Levedahl
@ 2026-05-15 16:06                     ` Johannes Sixt
  2026-05-16 14:38                       ` Mark Levedahl
  0 siblings, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-15 16:06 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 14.05.26 um 16:33 schrieb Mark Levedahl:
> git-gui attempts to use env(GIT_DIR) directly as the git repository,
> accepting GIT_DIR if it is a directory. Only if that fails is git
> rev-parse used to discover the repository.  But, this avoids all of
> git-core's validity checking on a repository, thus possibly deferring an
> error to a later step, possibly unexpected. Repository validation should
> be part of initial setup so that later processing does not need error
> trapping for configuration errors.

OK. If the user gave us GIT_DIR with our without GIT_WORK_TREE, then
that combination better be workable.

> 
> Let's just invoke rev-parse so all error checking is done. Stop here if
> the user set GIT_DIR or GIT_WORK_TREE. Otherwise, continue the existing
> behavior and show the repository picker.

OK. But the paragraph is confusing, because a big "If an error occurs"
is missing after the first sentence.

> 
> Also, remove a later check on whether _gitdir is a directory: that code
> cannot be reached without rev-parse having validating the repository.

Good.

> 
> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
> ---
>  git-gui.sh | 24 +++++++++---------------
>  1 file changed, 9 insertions(+), 15 deletions(-)
> 
> diff --git a/git-gui.sh b/git-gui.sh
> index 2e2ddc0..81789dd 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -374,6 +374,7 @@ set _gitdir {}
>  set _gitworktree {}
>  set _isbare {}
>  set _githtmldir {}
> +set _prefix {}
>  set _reponame {}
>  set _shellpath {@@SHELL_PATH@@}
>  
> @@ -1167,19 +1168,18 @@ proc pick_repo {} {
>  	set picked 1
>  }
>  
> +# find repository.
>  if {[catch {
> -		set _gitdir $env(GIT_DIR)
> -		set _prefix {}
> -		}]
> -	&& [catch {
> -		# beware that from the .git dir this sets _gitdir to .
> -		# and _prefix to the empty string
> -		set _gitdir [git rev-parse --absolute-git-dir]
> -		set _prefix [git rev-parse --show-prefix]
> -	} err]} {
> +	set _gitdir [git rev-parse --absolute-git-dir]

Please do also set _prefix. It should fix the bug that the file chooser
uses an empty prefix after

cd lib
GIT_DIR=$PWD/../.git GIT_WORK_TREE=$PWD/.. ../git-gui.sh browser master .

(this is an old bug.)

Please keep the additional indentation of the catch body.

> +} err]} {
> +	if {[is_gitvars_error $err]} {
> +		exit 1
> +	} else {
>  		pick_repo
> +	}

Treat the 'if' as an early exist without an else, and we don't need the
previously strange indentation of 'pick_repo'.

>  }
>  
> +
>  # Use object format as hash algorithm (either "sha1" or "sha256")
>  set hashalgorithm [git rev-parse --show-object-format]
>  if {$hashalgorithm eq "sha1"} {
> @@ -1191,12 +1191,6 @@ if {$hashalgorithm eq "sha1"} {
>  	exit 1
>  }
>  
> -if {![file isdirectory $_gitdir]} {
> -	catch {wm withdraw .}
> -	error_popup [strcat [mc "Git directory not found:"] "\n\n$_gitdir"]
> -	exit 1
> -}
> -
>  # _gitdir exists, so try loading the config
>  load_config 0
>  apply_config

(Stopping the review here for today.)

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 07/11] git-gui: use rev-parse exclusively to find a repository
  2026-05-15 16:06                     ` Johannes Sixt
@ 2026-05-16 14:38                       ` Mark Levedahl
  0 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-16 14:38 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: egg_mushroomcow, bootaina702, git



On 5/15/26 12:06 PM, Johannes Sixt wrote:
> Am 14.05.26 um 16:33 schrieb Mark Levedahl:
>> git-gui attempts to use env(GIT_DIR) directly as the git repository,
>> accepting GIT_DIR if it is a directory. Only if that fails is git
>> rev-parse used to discover the repository.  But, this avoids all of
>> git-core's validity checking on a repository, thus possibly deferring an
>> error to a later step, possibly unexpected. Repository validation should
>> be part of initial setup so that later processing does not need error
>> trapping for configuration errors.
> OK. If the user gave us GIT_DIR with our without GIT_WORK_TREE, then
> that combination better be workable.
>
>> Let's just invoke rev-parse so all error checking is done. Stop here if
>> the user set GIT_DIR or GIT_WORK_TREE. Otherwise, continue the existing
>> behavior and show the repository picker.
> OK. But the paragraph is confusing, because a big "If an error occurs"
> is missing after the first sentence.
will fix.
>> Also, remove a later check on whether _gitdir is a directory: that code
>> cannot be reached without rev-parse having validating the repository.
> Good.
>
>> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
>> ---
>>  git-gui.sh | 24 +++++++++---------------
>>  1 file changed, 9 insertions(+), 15 deletions(-)
>>
>> diff --git a/git-gui.sh b/git-gui.sh
>> index 2e2ddc0..81789dd 100755
>> --- a/git-gui.sh
>> +++ b/git-gui.sh
>> @@ -374,6 +374,7 @@ set _gitdir {}
>>  set _gitworktree {}
>>  set _isbare {}
>>  set _githtmldir {}
>> +set _prefix {}
>>  set _reponame {}
>>  set _shellpath {@@SHELL_PATH@@}
>>  
>> @@ -1167,19 +1168,18 @@ proc pick_repo {} {
>>  	set picked 1
>>  }
>>  
>> +# find repository.
>>  if {[catch {
>> -		set _gitdir $env(GIT_DIR)
>> -		set _prefix {}
>> -		}]
>> -	&& [catch {
>> -		# beware that from the .git dir this sets _gitdir to .
>> -		# and _prefix to the empty string
>> -		set _gitdir [git rev-parse --absolute-git-dir]
>> -		set _prefix [git rev-parse --show-prefix]
>> -	} err]} {
>> +	set _gitdir [git rev-parse --absolute-git-dir]
> Please do also set _prefix. It should fix the bug that the file chooser
> uses an empty prefix after
>
> cd lib
> GIT_DIR=$PWD/../.git GIT_WORK_TREE=$PWD/.. ../git-gui.sh browser master .
>
> (this is an old bug.)
>
> Please keep the additional indentation of the catch body.
>
>> +} err]} {
>> +	if {[is_gitvars_error $err]} {
>> +		exit 1
>> +	} else {
>>  		pick_repo
>> +	}
> Treat the 'if' as an early exist without an else, and we don't need the
> previously strange indentation of 'pick_repo'.
>
>>  }
>>  
>> +
>>  # Use object format as hash algorithm (either "sha1" or "sha256")
>>  set hashalgorithm [git rev-parse --show-object-format]
>>  if {$hashalgorithm eq "sha1"} {
>> @@ -1191,12 +1191,6 @@ if {$hashalgorithm eq "sha1"} {
>>  	exit 1
>>  }
>>  
>> -if {![file isdirectory $_gitdir]} {
>> -	catch {wm withdraw .}
>> -	error_popup [strcat [mc "Git directory not found:"] "\n\n$_gitdir"]
>> -	exit 1
>> -}
>> -
>>  # _gitdir exists, so try loading the config
>>  load_config 0
>>  apply_config

will fix all.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v1 08/11] git-gui: simplify [is_bare] to report if a worktree is known
  2026-05-14 14:33                 ` [PATCH v1 00/11] Improve git gui operation without a worktree Mark Levedahl
                                     ` (6 preceding siblings ...)
  2026-05-14 14:33                   ` [PATCH v1 07/11] git-gui: use rev-parse exclusively to find a repository Mark Levedahl
@ 2026-05-14 14:33                   ` Mark Levedahl
  2026-05-16  8:12                     ` Johannes Sixt
  2026-05-14 14:33                   ` [PATCH v1 09/11] git-gui: support using repository parent dir as a worktree Mark Levedahl
                                     ` (4 subsequent siblings)
  12 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-14 14:33 UTC (permalink / raw)
  To: git; +Cc: egg_mushroomcow, j6t, bootaina702, Mark Levedahl

git-gui includes proc is_bare, used in several places to make decisions
on whether a worktree exists, but also in discovery to tell if a
worktree can be supported.

But, is_bare is out of date with regard to multiple worktrees, safe
repository guards, and possibly other relevant features known to git
rev-parse. Also, is_bare caches its result on the first call, so is not
useful if a later step in the discovery process finds a worktree.

So, simplify is_bare to report whether git-gui has a worktree or is
working only from a repository.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 25 +------------------------
 1 file changed, 1 insertion(+), 24 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 81789dd..a03eaa7 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -372,7 +372,6 @@ if {[tk windowingsystem] eq "aqua"} {
 set _appname {Git Gui}
 set _gitdir {}
 set _gitworktree {}
-set _isbare {}
 set _githtmldir {}
 set _prefix {}
 set _reponame {}
@@ -524,29 +523,7 @@ proc get_config {name} {
 }
 
 proc is_bare {} {
-	global _isbare
-	global _gitdir
-	global _gitworktree
-
-	if {$_isbare eq {}} {
-		if {[catch {
-			set _bare [git rev-parse --is-bare-repository]
-			switch  -- $_bare {
-			true { set _isbare 1 }
-			false { set _isbare 0}
-			default { throw }
-			}
-		}]} {
-			if {[is_config_true core.bare]
-				|| ($_gitworktree eq {}
-					&& [lindex [file split $_gitdir] end] ne {.git})} {
-				set _isbare 1
-			} else {
-				set _isbare 0
-			}
-		}
-	}
-	return $_isbare
+	return [expr {$::_gitworktree eq {}}]
 }
 
 ######################################################################
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 08/11] git-gui: simplify [is_bare] to report if a worktree is known
  2026-05-14 14:33                   ` [PATCH v1 08/11] git-gui: simplify [is_bare] to report if a worktree is known Mark Levedahl
@ 2026-05-16  8:12                     ` Johannes Sixt
  0 siblings, 0 replies; 134+ messages in thread
From: Johannes Sixt @ 2026-05-16  8:12 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 14.05.26 um 16:33 schrieb Mark Levedahl:
> git-gui includes proc is_bare, used in several places to make decisions
> on whether a worktree exists, but also in discovery to tell if a
> worktree can be supported.
> 
> But, is_bare is out of date with regard to multiple worktrees, safe
> repository guards, and possibly other relevant features known to git
> rev-parse. Also, is_bare caches its result on the first call, so is not
> useful if a later step in the discovery process finds a worktree.
> 
> So, simplify is_bare to report whether git-gui has a worktree or is
> working only from a repository.
> 
> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
> ---
>  git-gui.sh | 25 +------------------------
>  1 file changed, 1 insertion(+), 24 deletions(-)
> 
> diff --git a/git-gui.sh b/git-gui.sh
> index 81789dd..a03eaa7 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -372,7 +372,6 @@ if {[tk windowingsystem] eq "aqua"} {
>  set _appname {Git Gui}
>  set _gitdir {}
>  set _gitworktree {}
> -set _isbare {}
>  set _githtmldir {}
>  set _prefix {}
>  set _reponame {}
> @@ -524,29 +523,7 @@ proc get_config {name} {
>  }
>  
>  proc is_bare {} {
> -	global _isbare
> -	global _gitdir
> -	global _gitworktree
> -
> -	if {$_isbare eq {}} {
> -		if {[catch {
> -			set _bare [git rev-parse --is-bare-repository]
> -			switch  -- $_bare {
> -			true { set _isbare 1 }
> -			false { set _isbare 0}
> -			default { throw }
> -			}
> -		}]} {
> -			if {[is_config_true core.bare]
> -				|| ($_gitworktree eq {}
> -					&& [lindex [file split $_gitdir] end] ne {.git})} {
> -				set _isbare 1
> -			} else {
> -				set _isbare 0
> -			}
> -		}
> -	}
> -	return $_isbare
> +	return [expr {$::_gitworktree eq {}}]
>  }
>  
>  ######################################################################

Very nice!

IMO, regardless of which way we end up rewriting repository discovery,
the end result should be that we can use $_gitworktree like this to tell
whether we are in a bare repository or not.

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v1 09/11] git-gui: support using repository parent dir as a worktree
  2026-05-14 14:33                 ` [PATCH v1 00/11] Improve git gui operation without a worktree Mark Levedahl
                                     ` (7 preceding siblings ...)
  2026-05-14 14:33                   ` [PATCH v1 08/11] git-gui: simplify [is_bare] to report if a worktree is known Mark Levedahl
@ 2026-05-14 14:33                   ` Mark Levedahl
  2026-05-16  8:14                     ` Johannes Sixt
  2026-05-14 14:33                   ` [PATCH v1 10/11] git-gui: improve worktree discovery Mark Levedahl
                                     ` (3 subsequent siblings)
  12 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-14 14:33 UTC (permalink / raw)
  To: git; +Cc: egg_mushroomcow, j6t, bootaina702, Mark Levedahl

git-gui, since 87cd09f43e ("git-gui: work from the .git dir",
2010-01-23), has had the intent to allow starting from inside a
repository, then switching to the parent directory if that is a valid
worktree.

This certainly hasn't worked since 2d92ab32fd ("rev-parse: make
--show-toplevel without a worktree an error", 2019-11-19) in git, but
breaking this git-gui feature was unintentional.

Add a proc to test if the parent of the git repository is a valid
worktree, and set that directory as the worktree if so. Use invocations
of git rev-parse to assure all validity and safety checks included in
git-core are executed.
---
 git-gui.sh | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

diff --git a/git-gui.sh b/git-gui.sh
index a03eaa7..e326401 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1100,6 +1100,23 @@ unset argv0dir
 ##
 ## repository setup
 
+proc is_parent_worktree {} {
+	# Directory 'parent' of a repository named 'parent/.git' might be the worktree
+	set ok 0
+	if {[file tail $::_gitdir] eq {.git}} {
+		set gitdir_parent [file join $::_gitdir {..}]
+		set expected_worktree [file normalize $gitdir_parent]
+		catch {set git_worktree [git -C $gitdir_parent rev-parse --show-toplevel]}
+		if {[string compare $expected_worktree $git_worktree] == 0} {
+			set ::_prefix {}
+			set ::_gitworktree $git_worktree
+			cd $git_worktree
+			set ok 1
+		}
+	}
+	return $ok
+}
+
 proc is_gitvars_error {err} {
 	set havevars 0
 	set GIT_DIR {}
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 09/11] git-gui: support using repository parent dir as a worktree
  2026-05-14 14:33                   ` [PATCH v1 09/11] git-gui: support using repository parent dir as a worktree Mark Levedahl
@ 2026-05-16  8:14                     ` Johannes Sixt
  2026-05-16 14:48                       ` Mark Levedahl
  0 siblings, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-16  8:14 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 14.05.26 um 16:33 schrieb Mark Levedahl:
> git-gui, since 87cd09f43e ("git-gui: work from the .git dir",
> 2010-01-23), has had the intent to allow starting from inside a
> repository, then switching to the parent directory if that is a valid
> worktree.
> 
> This certainly hasn't worked since 2d92ab32fd ("rev-parse: make
> --show-toplevel without a worktree an error", 2019-11-19) in git, but
> breaking this git-gui feature was unintentional.
> 
> Add a proc to test if the parent of the git repository is a valid
> worktree, and set that directory as the worktree if so. Use invocations
> of git rev-parse to assure all validity and safety checks included in
> git-core are executed.

BTW, missing sign-off.

> ---
>  git-gui.sh | 17 +++++++++++++++++
>  1 file changed, 17 insertions(+)
> 
> diff --git a/git-gui.sh b/git-gui.sh
> index a03eaa7..e326401 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -1100,6 +1100,23 @@ unset argv0dir
>  ##
>  ## repository setup
>  
> +proc is_parent_worktree {} {
> +	# Directory 'parent' of a repository named 'parent/.git' might be the worktree
> +	set ok 0
> +	if {[file tail $::_gitdir] eq {.git}} {
> +		set gitdir_parent [file join $::_gitdir {..}]
> +		set expected_worktree [file normalize $gitdir_parent]

We have [file dirname ...]. Is there a reason to avoid it?

> +		catch {set git_worktree [git -C $gitdir_parent rev-parse --show-toplevel]}
> +		if {[string compare $expected_worktree $git_worktree] == 0} {

The purpose of this check should be explained in a comment. I think it is:

For a repository with the database in a directory named .git we assume
that the working tree is the directory containing .git. But
configuration may point to a different worktree. Then we do not want to
hold on to our assumption.

However, whether [git -C elsewhere ...] uses the same gitdir that we
have discovered so far cannot be told from this piece of code alone.
Therefore, I think it is wrong to extract this check into a function.

Also, I don't think we can use string comparison here. On Windows, the
command returns the Windows style path, but Tcl my operate with a POSIX
style path.

> +			set ::_prefix {}
> +			set ::_gitworktree $git_worktree
> +			cd $git_worktree

So many side-effects in a function whose name suggests that it only does
some checks. Please, don't do that.

> +			set ok 1
> +		}
> +	}
> +	return $ok
> +}
> +
>  proc is_gitvars_error {err} {
>  	set havevars 0
>  	set GIT_DIR {}

In general, I am not a fan of commits that add new functions, but no
call sites. Please squash this into 10/11. Ditto for is_gitvars_error in
06/11.

-- Hannes

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 09/11] git-gui: support using repository parent dir as a worktree
  2026-05-16  8:14                     ` Johannes Sixt
@ 2026-05-16 14:48                       ` Mark Levedahl
  0 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-16 14:48 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: egg_mushroomcow, bootaina702, git



On 5/16/26 4:14 AM, Johannes Sixt wrote:
> Am 14.05.26 um 16:33 schrieb Mark Levedahl:
>> git-gui, since 87cd09f43e ("git-gui: work from the .git dir",
>> 2010-01-23), has had the intent to allow starting from inside a
>> repository, then switching to the parent directory if that is a valid
>> worktree.
>>
>> This certainly hasn't worked since 2d92ab32fd ("rev-parse: make
>> --show-toplevel without a worktree an error", 2019-11-19) in git, but
>> breaking this git-gui feature was unintentional.
>>
>> Add a proc to test if the parent of the git repository is a valid
>> worktree, and set that directory as the worktree if so. Use invocations
>> of git rev-parse to assure all validity and safety checks included in
>> git-core are executed.
> BTW, missing sign-off.
>
>> ---
>>  git-gui.sh | 17 +++++++++++++++++
>>  1 file changed, 17 insertions(+)
>>
>> diff --git a/git-gui.sh b/git-gui.sh
>> index a03eaa7..e326401 100755
>> --- a/git-gui.sh
>> +++ b/git-gui.sh
>> @@ -1100,6 +1100,23 @@ unset argv0dir
>>  ##
>>  ## repository setup
>>  
>> +proc is_parent_worktree {} {
>> +	# Directory 'parent' of a repository named 'parent/.git' might be the worktree
>> +	set ok 0
>> +	if {[file tail $::_gitdir] eq {.git}} {
>> +		set gitdir_parent [file join $::_gitdir {..}]
>> +		set expected_worktree [file normalize $gitdir_parent]
> We have [file dirname ...]. Is there a reason to avoid it?
>
>> +		catch {set git_worktree [git -C $gitdir_parent rev-parse --show-toplevel]}
>> +		if {[string compare $expected_worktree $git_worktree] == 0} {
> The purpose of this check should be explained in a comment. I think it is:
>
> For a repository with the database in a directory named .git we assume
> that the working tree is the directory containing .git. But
> configuration may point to a different worktree. Then we do not want to
> hold on to our assumption.
>
> However, whether [git -C elsewhere ...] uses the same gitdir that we
> have discovered so far cannot be told from this piece of code alone.
> Therefore, I think it is wrong to extract this check into a function.
>
> Also, I don't think we can use string comparison here. On Windows, the
> command returns the Windows style path, but Tcl my operate with a POSIX
> style path.
As you have correctly inferred, am trying to unambiguously establish that git running in
the parent directory is using  the child .git as the repository. I think this actually
requires two calls to git-revparse (--absolute-git-dir and --show-toplevel).
- the current git repo is valid to support a worktree. Will rework.

>> +			set ::_prefix {}
>> +			set ::_gitworktree $git_worktree
>> +			cd $git_worktree
> So many side-effects in a function whose name suggests that it only does
> some checks. Please, don't do that.
>
>> +			set ok 1
>> +		}
>> +	}
>> +	return $ok
>> +}
>> +
>>  proc is_gitvars_error {err} {
>>  	set havevars 0
>>  	set GIT_DIR {}
> In general, I am not a fan of commits that add new functions, but no
> call sites. Please squash this into 10/11. Ditto for is_gitvars_error in
> 06/11.
>
> -- Hannes
>

Next round should address all of your comments.
Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v1 10/11] git-gui: improve worktree discovery
  2026-05-14 14:33                 ` [PATCH v1 00/11] Improve git gui operation without a worktree Mark Levedahl
                                     ` (8 preceding siblings ...)
  2026-05-14 14:33                   ` [PATCH v1 09/11] git-gui: support using repository parent dir as a worktree Mark Levedahl
@ 2026-05-14 14:33                   ` Mark Levedahl
  2026-05-16  8:16                     ` Johannes Sixt
  2026-05-14 14:33                   ` [PATCH v1 11/11] git-gui: add gui and pick as explicit subcommands Mark Levedahl
                                     ` (2 subsequent siblings)
  12 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-14 14:33 UTC (permalink / raw)
  To: git; +Cc: egg_mushroomcow, j6t, bootaina702, Mark Levedahl

git gui's worktree discovery needs update based upon prior work in this
series. In the normal case, all information we need comes directly from
git rev-parse (--show-toplevel, and --show-prefix). Should this work, we
have a valid worktree and all git gui commands can run.

If not, we need to consider:
- if GIT_DIR or GIT_WORK_TREE are in the environment, just stop as we
  the input configuration was wrong, the user must fix that.
- if we have a browser or blame subcommand, no worktree is needed so
  git-gui can run without.
- using the git repository's parent is a valid worktree (if possible),
  restoring prior behavior.

The current directory should be either the root of the worktree, if one
is found, or the top-level of the git repository.

Make it so. Also, make worktree discover directly follow repository
discovery, reducing the locations that might need error trapping to
catch configuration issues.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 56 ++++++++++++++++++++++--------------------------------
 1 file changed, 23 insertions(+), 33 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index e326401..3a83dd5 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1173,6 +1173,28 @@ if {[catch {
 	}
 }
 
+# find worktree, continue without if not required
+if {[catch {
+	set _gitworktree [git rev-parse --show-toplevel]
+	set _prefix [git rev-parse --show-prefix]
+	cd $_gitworktree
+} err]} {
+	if {[is_gitvars_error $err]} {
+		exit 1
+	}
+	set _gitworktree {}
+	set _prefix {}
+	if {[is_enabled bare]} {
+		cd $_gitdir
+	} elseif {![is_parent_worktree]} {
+		catch {wm withdraw .}
+		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n" $_gitdir]
+		exit 1
+	}
+}
+
+# repository and worktree config are complete, export them
+set_gitdir_vars
 
 # Use object format as hash algorithm (either "sha1" or "sha256")
 set hashalgorithm [git rev-parse --show-object-format]
@@ -1189,37 +1211,8 @@ if {$hashalgorithm eq "sha1"} {
 load_config 0
 apply_config
 
-set _gitworktree [git rev-parse --show-toplevel]
 
-if {$_prefix ne {}} {
-	if {$_gitworktree eq {}} {
-		regsub -all {[^/]+/} $_prefix ../ cdup
-	} else {
-		set cdup $_gitworktree
-	}
-	if {[catch {cd $cdup} err]} {
-		catch {wm withdraw .}
-		error_popup [strcat [mc "Cannot move to top of working directory:"] "\n\n$err"]
-		exit 1
-	}
-	set _gitworktree [pwd]
-	unset cdup
-} elseif {![is_enabled bare]} {
-	if {[is_bare]} {
-		catch {wm withdraw .}
-		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n$_gitdir"]
-		exit 1
-	}
-	if {$_gitworktree eq {}} {
-		set _gitworktree [file dirname $_gitdir]
-	}
-	if {[catch {cd $_gitworktree} err]} {
-		catch {wm withdraw .}
-		error_popup [strcat [mc "No working directory"] " $_gitworktree:\n\n$err"]
-		exit 1
-	}
-	set _gitworktree [pwd]
-}
+# Derive a human-readable repository name
 set _reponame [file split [file normalize $_gitdir]]
 if {[lindex $_reponame end] eq {.git}} {
 	set _reponame [lindex $_reponame end-1]
@@ -1227,9 +1220,6 @@ if {[lindex $_reponame end] eq {.git}} {
 	set _reponame [lindex $_reponame end]
 }
 
-# Export the final paths
-set_gitdir_vars
-
 ######################################################################
 ##
 ## global init
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 10/11] git-gui: improve worktree discovery
  2026-05-14 14:33                   ` [PATCH v1 10/11] git-gui: improve worktree discovery Mark Levedahl
@ 2026-05-16  8:16                     ` Johannes Sixt
  2026-05-16 15:28                       ` Mark Levedahl
  0 siblings, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-16  8:16 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 14.05.26 um 16:33 schrieb Mark Levedahl:
> git gui's worktree discovery needs update based upon prior work in this
> series. In the normal case, all information we need comes directly from
> git rev-parse (--show-toplevel, and --show-prefix). Should this work, we
> have a valid worktree and all git gui commands can run.
> 
> If not, we need to consider:
> - if GIT_DIR or GIT_WORK_TREE are in the environment, just stop as we
>   the input configuration was wrong, the user must fix that.
> - if we have a browser or blame subcommand, no worktree is needed so
>   git-gui can run without.
> - using the git repository's parent is a valid worktree (if possible),
>   restoring prior behavior.
> 
> The current directory should be either the root of the worktree, if one
> is found, or the top-level of the git repository.

I disagree in the case where no working tree is found. Then there is no
point in changing the current directory.

> 
> Make it so. Also, make worktree discover directly follow repository
> discovery, reducing the locations that might need error trapping to
> catch configuration issues.

Good!

> 
> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>

BTW, please make the subject line more descriptive. The word "improve"
does not convey anything of importance.

> ---
>  git-gui.sh | 56 ++++++++++++++++++++++--------------------------------
>  1 file changed, 23 insertions(+), 33 deletions(-)
> 
> diff --git a/git-gui.sh b/git-gui.sh
> index e326401..3a83dd5 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -1173,6 +1173,28 @@ if {[catch {
>  	}
>  }
>  
> +# find worktree, continue without if not required
> +if {[catch {
> +	set _gitworktree [git rev-parse --show-toplevel]
> +	set _prefix [git rev-parse --show-prefix]
> +	cd $_gitworktree
> +} err]} {

We have three commands, each with their own possible failure sources.
One of the outcomes of an error is that we proceed anyway. I think that
this is incorrect if the 'cd' fails: we must not proceed if it fails.
Therefore, we must handle its failure separately.

> +	if {[is_gitvars_error $err]} {
> +		exit 1
> +	}
> +	set _gitworktree {}
> +	set _prefix {}
> +	if {[is_enabled bare]} {
> +		cd $_gitdir

Why change the directory here? If we run `git gui browser master dir` we
do not want to change the directory in an uncontrolled manner. The
argument parser will want to check for the existence of files, and then
we do not want to operate from a random directory.

Also, I think that the check must be for [is_bare] and not [is_enabled
bare].

> +	} elseif {![is_parent_worktree]} {
> +		catch {wm withdraw .}
> +		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n" $_gitdir]
> +		exit 1
> +	}
> +}
> +
> +# repository and worktree config are complete, export them
> +set_gitdir_vars
>  
>  # Use object format as hash algorithm (either "sha1" or "sha256")
>  set hashalgorithm [git rev-parse --show-object-format]

This moves code around. In particular, we see load_config and
apply_config in the context below, which now happens only after these
calls. How certain are we that these have no effect on the code that
runs now earlier?

> @@ -1189,37 +1211,8 @@ if {$hashalgorithm eq "sha1"} {
>  load_config 0
>  apply_config
>  
> -set _gitworktree [git rev-parse --show-toplevel]
>  
> -if {$_prefix ne {}} {
> -	if {$_gitworktree eq {}} {
> -		regsub -all {[^/]+/} $_prefix ../ cdup
> -	} else {
> -		set cdup $_gitworktree
> -	}
> -	if {[catch {cd $cdup} err]} {
> -		catch {wm withdraw .}
> -		error_popup [strcat [mc "Cannot move to top of working directory:"] "\n\n$err"]
> -		exit 1
> -	}
> -	set _gitworktree [pwd]
> -	unset cdup
> -} elseif {![is_enabled bare]} {
> -	if {[is_bare]} {
> -		catch {wm withdraw .}
> -		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n$_gitdir"]
> -		exit 1
> -	}
> -	if {$_gitworktree eq {}} {
> -		set _gitworktree [file dirname $_gitdir]
> -	}
> -	if {[catch {cd $_gitworktree} err]} {
> -		catch {wm withdraw .}
> -		error_popup [strcat [mc "No working directory"] " $_gitworktree:\n\n$err"]
> -		exit 1
> -	}
> -	set _gitworktree [pwd]
> -}
> +# Derive a human-readable repository name
>  set _reponame [file split [file normalize $_gitdir]]
>  if {[lindex $_reponame end] eq {.git}} {
>  	set _reponame [lindex $_reponame end-1]
> @@ -1227,9 +1220,6 @@ if {[lindex $_reponame end] eq {.git}} {
>  	set _reponame [lindex $_reponame end]
>  }
>  
> -# Export the final paths
> -set_gitdir_vars
> -
>  ######################################################################
>  ##
>  ## global init

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 10/11] git-gui: improve worktree discovery
  2026-05-16  8:16                     ` Johannes Sixt
@ 2026-05-16 15:28                       ` Mark Levedahl
  2026-05-19  8:16                         ` Johannes Sixt
  0 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-16 15:28 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: egg_mushroomcow, bootaina702, git



On 5/16/26 4:16 AM, Johannes Sixt wrote:
> Am 14.05.26 um 16:33 schrieb Mark Levedahl:
>> git gui's worktree discovery needs update based upon prior work in this
>> series. In the normal case, all information we need comes directly from
>> git rev-parse (--show-toplevel, and --show-prefix). Should this work, we
>> have a valid worktree and all git gui commands can run.
>>
>> If not, we need to consider:
>> - if GIT_DIR or GIT_WORK_TREE are in the environment, just stop as we
>>   the input configuration was wrong, the user must fix that.
>> - if we have a browser or blame subcommand, no worktree is needed so
>>   git-gui can run without.
>> - using the git repository's parent is a valid worktree (if possible),
>>   restoring prior behavior.
>>
>> The current directory should be either the root of the worktree, if one
>> is found, or the top-level of the git repository.
> I disagree in the case where no working tree is found. Then there is no
> point in changing the current directory.

git-gui always changes to root of the worktree if that is found.
Failing to cd to the root of the repository when operating with no worktree opens the
possibility th
>
>> Make it so. Also, make worktree discover directly follow repository
>> discovery, reducing the locations that might need error trapping to
>> catch configuration issues.
> Good!
>
>> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
> BTW, please make the subject line more descriptive. The word "improve"
> does not convey anything of importance.
>
>> ---
>>  git-gui.sh | 56 ++++++++++++++++++++++--------------------------------
>>  1 file changed, 23 insertions(+), 33 deletions(-)
>>
>> diff --git a/git-gui.sh b/git-gui.sh
>> index e326401..3a83dd5 100755
>> --- a/git-gui.sh
>> +++ b/git-gui.sh
>> @@ -1173,6 +1173,28 @@ if {[catch {
>>  	}
>>  }
>>  
>> +# find worktree, continue without if not required
>> +if {[catch {
>> +	set _gitworktree [git rev-parse --show-toplevel]
>> +	set _prefix [git rev-parse --show-prefix]
>> +	cd $_gitworktree
>> +} err]} {
> We have three commands, each with their own possible failure sources.
> One of the outcomes of an error is that we proceed anyway. I think that
> this is incorrect if the 'cd' fails: we must not proceed if it fails.
> Therefore, we must handle its failure separately.
>
>> +	if {[is_gitvars_error $err]} {
>> +		exit 1
>> +	}
>> +	set _gitworktree {}
>> +	set _prefix {}
>> +	if {[is_enabled bare]} {
>> +		cd $_gitdir
> Why change the directory here? If we run `git gui browser master dir` we
> do not want to change the directory in an uncontrolled manner. The
> argument parser will want to check for the existence of files, and then
> we do not want to operate from a random directory.
>
> Also, I think that the check must be for [is_bare] and not [is_enabled
> bare].

[is_enabled_bare] is correct. This code handles the case: 
    - neither the startup directory nor GIT_WORK_TREE are useable worktrees, so [is_bare]
is currently true.
    - the command given is browser or blame so a worktree is not needed. We can proceed.

The bigger question is whether to change directory at all: git-gui should never touch
files that are neither in the worktree nor in the repository. Leaving the current
directory as neither of those could be troublesome. I have no strong feeling here though,
will delete this.
    

>> +	} elseif {![is_parent_worktree]} {
>> +		catch {wm withdraw .}
>> +		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n" $_gitdir]
>> +		exit 1
>> +	}
>> +}
>> +
>> +# repository and worktree config are complete, export them
>> +set_gitdir_vars
>>  
>>  # Use object format as hash algorithm (either "sha1" or "sha256")
>>  set hashalgorithm [git rev-parse --show-object-format]
> This moves code around. In particular, we see load_config and
> apply_config in the context below, which now happens only after these
> calls. How certain are we that these have no effect on the code that
> runs now earlier?
We need to load the system and user global config before running the repository picker. We
(re-) load the full config including the repository after we have a repository. I think
this is correct: git-config explicitly lists worktree dependent includeif statements,
meaning the config can be worktree dependent, and we must not load the final config until
repository and worktree discovery are complete.

git rev-parse, etc., perform discovery and config file loading each time they are invoked,
those are unaffected by git-gui's internal config.

I will clarify this explicitly in the commit message.

>> @@ -1189,37 +1211,8 @@ if {$hashalgorithm eq "sha1"} {
>>  load_config 0
>>  apply_config
>>  
>> -set _gitworktree [git rev-parse --show-toplevel]
>>  
>> -if {$_prefix ne {}} {
>> -	if {$_gitworktree eq {}} {
>> -		regsub -all {[^/]+/} $_prefix ../ cdup
>> -	} else {
>> -		set cdup $_gitworktree
>> -	}
>> -	if {[catch {cd $cdup} err]} {
>> -		catch {wm withdraw .}
>> -		error_popup [strcat [mc "Cannot move to top of working directory:"] "\n\n$err"]
>> -		exit 1
>> -	}
>> -	set _gitworktree [pwd]
>> -	unset cdup
>> -} elseif {![is_enabled bare]} {
>> -	if {[is_bare]} {
>> -		catch {wm withdraw .}
>> -		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n$_gitdir"]
>> -		exit 1
>> -	}
>> -	if {$_gitworktree eq {}} {
>> -		set _gitworktree [file dirname $_gitdir]
>> -	}
>> -	if {[catch {cd $_gitworktree} err]} {
>> -		catch {wm withdraw .}
>> -		error_popup [strcat [mc "No working directory"] " $_gitworktree:\n\n$err"]
>> -		exit 1
>> -	}
>> -	set _gitworktree [pwd]
>> -}
>> +# Derive a human-readable repository name
>>  set _reponame [file split [file normalize $_gitdir]]
>>  if {[lindex $_reponame end] eq {.git}} {
>>  	set _reponame [lindex $_reponame end-1]
>> @@ -1227,9 +1220,6 @@ if {[lindex $_reponame end] eq {.git}} {
>>  	set _reponame [lindex $_reponame end]
>>  }
>>  
>> -# Export the final paths
>> -set_gitdir_vars
>> -
>>  ######################################################################
>>  ##
>>  ## global init
> -- Hannes
>


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 10/11] git-gui: improve worktree discovery
  2026-05-16 15:28                       ` Mark Levedahl
@ 2026-05-19  8:16                         ` Johannes Sixt
  2026-05-19 19:00                           ` Mark Levedahl
  0 siblings, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-19  8:16 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 16.05.26 um 17:28 schrieb Mark Levedahl:
> On 5/16/26 4:16 AM, Johannes Sixt wrote:
>> Am 14.05.26 um 16:33 schrieb Mark Levedahl:
>>> +	if {[is_gitvars_error $err]} {
>>> +		exit 1
>>> +	}
>>> +	set _gitworktree {}
>>> +	set _prefix {}
>>> +	if {[is_enabled bare]} {
>>> +		cd $_gitdir
>> Why change the directory here? If we run `git gui browser master dir` we
>> do not want to change the directory in an uncontrolled manner. The
>> argument parser will want to check for the existence of files, and then
>> we do not want to operate from a random directory.
>>
>> Also, I think that the check must be for [is_bare] and not [is_enabled
>> bare].
> 
> [is_enabled_bare] is correct. This code handles the case: 
>     - neither the startup directory nor GIT_WORK_TREE are useable worktrees, so [is_bare]
> is currently true.
>     - the command given is browser or blame so a worktree is not needed. We can proceed.

But in the case where the command is browser or blame, the argument
parser must later check for the existence of files, provided that a
worktree is present. But this conditional would change directory to
somewhere that is not a worktree at all even though a worktree is
available. So, I am still convinced that [is_bare] is correct.

> The bigger question is whether to change directory at all: git-gui should never touch
> files that are neither in the worktree nor in the repository. Leaving the current
> directory as neither of those could be troublesome. I have no strong feeling here though,
> will delete this.

OK. We need the conditional, but not change the directory.

>>> +	} elseif {![is_parent_worktree]} {
>>> +		catch {wm withdraw .}
>>> +		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n" $_gitdir]
>>> +		exit 1
>>> +	}
>>> +}
>>> +
>>> +# repository and worktree config are complete, export them
>>> +set_gitdir_vars
>>>  
>>>  # Use object format as hash algorithm (either "sha1" or "sha256")
>>>  set hashalgorithm [git rev-parse --show-object-format]
>> This moves code around. In particular, we see load_config and
>> apply_config in the context below, which now happens only after these
>> calls. How certain are we that these have no effect on the code that
>> runs now earlier?
> We need to load the system and user global config before running the repository picker. We
> (re-) load the full config including the repository after we have a repository. I think
> this is correct: git-config explicitly lists worktree dependent includeif statements,
> meaning the config can be worktree dependent, and we must not load the final config until
> repository and worktree discovery are complete.

Good point.

> 
> git rev-parse, etc., perform discovery and config file loading each time they are invoked,
> those are unaffected by git-gui's internal config.

Also very true. Only calls to proc get_config would be affected by a
different setup, but there are none in the code that has swapped places
with working tree discovery. Only --show-object-format would be affected
by the discovered environment, and for that case it is more correct to
operate in the final setup.

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 10/11] git-gui: improve worktree discovery
  2026-05-19  8:16                         ` Johannes Sixt
@ 2026-05-19 19:00                           ` Mark Levedahl
  0 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-19 19:00 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: egg_mushroomcow, bootaina702, git



On 5/19/26 4:16 AM, Johannes Sixt wrote:
> Am 16.05.26 um 17:28 schrieb Mark Levedahl:
>> On 5/16/26 4:16 AM, Johannes Sixt wrote:
>>> Am 14.05.26 um 16:33 schrieb Mark Levedahl:
>>>> +	if {[is_gitvars_error $err]} {
>>>> +		exit 1
>>>> +	}
>>>> +	set _gitworktree {}
>>>> +	set _prefix {}
>>>> +	if {[is_enabled bare]} {
>>>> +		cd $_gitdir
>>> Why change the directory here? If we run `git gui browser master dir` we
>>> do not want to change the directory in an uncontrolled manner. The
>>> argument parser will want to check for the existence of files, and then
>>> we do not want to operate from a random directory.
>>>
>>> Also, I think that the check must be for [is_bare] and not [is_enabled
>>> bare].
>> [is_enabled_bare] is correct. This code handles the case: 
>>     - neither the startup directory nor GIT_WORK_TREE are useable worktrees, so [is_bare]
>> is currently true.
>>     - the command given is browser or blame so a worktree is not needed. We can proceed.
> But in the case where the command is browser or blame, the argument
> parser must later check for the existence of files, provided that a
> worktree is present. But this conditional would change directory to
> somewhere that is not a worktree at all even though a worktree is
> available. So, I am still convinced that [is_bare] is correct.
I did change this.   but...
I have reworked the blame/browser parser so it fully matches git blame parsing for the
single rev + path (or path rev) cases, all now do the same thing with or without a
worktree as they all work from git history, so having a worktree becomes almost moot (I
found issues with how git-gui handles --, it is dead wrong in some cases if the intent is
to match git-blame). What doesn't change is what blame displays based upon uncommitted or
staged changes in the worktree (if it does anything, but comments from 2007 suggest it
does, I haven't tested). I've done nothing to change that, only finding the args to pass
in to browser / blame. So, if a worktree exists, blame will still use the info there as it
does now.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v1 11/11] git-gui: add gui and pick as explicit subcommands
  2026-05-14 14:33                 ` [PATCH v1 00/11] Improve git gui operation without a worktree Mark Levedahl
                                     ` (9 preceding siblings ...)
  2026-05-14 14:33                   ` [PATCH v1 10/11] git-gui: improve worktree discovery Mark Levedahl
@ 2026-05-14 14:33                   ` Mark Levedahl
  2026-05-16  8:18                     ` Johannes Sixt
  2026-05-16  8:28                   ` [PATCH v1 00/11] Improve git gui operation without a worktree Johannes Sixt
  2026-05-20 20:23                   ` [PATCH v2 " Mark Levedahl
  12 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-14 14:33 UTC (permalink / raw)
  To: git; +Cc: egg_mushroomcow, j6t, bootaina702, Mark Levedahl

git-gui accepts subcommands blame | browser | citool, and assumes the
subcommand is 'gui' if none is actually given, But, git gui also has a
repository picker (choose_repository::pick) that can create a new
repository + worktree, or choose an existing one, switch to that, and
the run the gui. The user has no direct control over invoking the
picker, instead the picker is triggered by failure in the repository /
worktree discover process: this includes being started in a directory
not controlled by git, which is probably the intended use case.

The picker can appear when the user has no intention of creating a new
worktree, and the user cannot use the picker to create a new worktree
inside another.

So, add two new explicit subcommands:
    gui  - Run the gui if repository/worktree discovery succeeds, or die
           with an error message, but never run the picker.
    pick - First run the picker, regardless, then start the gui in
           the chosen worktree.

Nothing in this changes the prior behavior, the alternates above must be
explicitly selected to see any change.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 22 ++++++++++++++++++++--
 1 file changed, 20 insertions(+), 2 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 3a83dd5..c56aeef 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1021,6 +1021,7 @@ proc load_config {include_global} {
 ##
 ## feature option selection
 
+set run_picker_on_error 1
 if {[regexp {^git-(.+)$} [file tail $argv0] _junk subcommand]} {
 	unset _junk
 } else {
@@ -1030,6 +1031,7 @@ if {$subcommand eq {gui.sh}} {
 	set subcommand gui
 }
 if {$subcommand eq {gui} && [llength $argv] > 0} {
+	set run_picker_on_error 0
 	set subcommand [lindex $argv 0]
 	set argv [lrange $argv 1 end]
 }
@@ -1047,6 +1049,7 @@ blame {
 	disable_option multicommit
 	disable_option branch
 	disable_option transport
+	set run_picker_on_error 0
 }
 citool {
 	enable_option singlecommit
@@ -1055,6 +1058,7 @@ citool {
 	disable_option multicommit
 	disable_option branch
 	disable_option transport
+	set run_picker_on_error 0
 
 	while {[llength $argv] > 0} {
 		set a [lindex $argv 0]
@@ -1162,14 +1166,28 @@ proc pick_repo {} {
 	set picked 1
 }
 
+# run repository picker if explicitly requested
+switch -- $subcommand {
+	pick {
+		pick_repo
+		set subcommand gui
+		set run_picker_on_error 0
+	}
+}
+
 # find repository.
 if {[catch {
 	set _gitdir [git rev-parse --absolute-git-dir]
 } err]} {
 	if {[is_gitvars_error $err]} {
 		exit 1
-	} else {
+	}
+	if {$run_picker_on_error} {
 		pick_repo
+	} else {
+		catch {wm withdraw .}
+		error_popup [strcat [mc "Git directory not found:"] "\n\n$err"]
+		exit 1
 	}
 }
 
@@ -3051,7 +3069,7 @@ gui {
 	# fall through to setup UI for commits
 }
 default {
-	set err "[mc usage:] $argv0 \[{blame|browser|citool}\]"
+	set err "[mc usage:] $argv0 \[{blame|browser|citool|gui|pick}\]"
 	if {[tk windowingsystem] eq "win32"} {
 		wm withdraw .
 		tk_messageBox -icon error -message $err \
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 11/11] git-gui: add gui and pick as explicit subcommands
  2026-05-14 14:33                   ` [PATCH v1 11/11] git-gui: add gui and pick as explicit subcommands Mark Levedahl
@ 2026-05-16  8:18                     ` Johannes Sixt
  2026-05-16 15:42                       ` Mark Levedahl
  0 siblings, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-16  8:18 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 14.05.26 um 16:33 schrieb Mark Levedahl:
> git-gui accepts subcommands blame | browser | citool, and assumes the
> subcommand is 'gui' if none is actually given, But, git gui also has a
> repository picker (choose_repository::pick) that can create a new
> repository + worktree, or choose an existing one, switch to that, and
> the run the gui. The user has no direct control over invoking the
> picker, instead the picker is triggered by failure in the repository /
> worktree discover process: this includes being started in a directory
> not controlled by git, which is probably the intended use case.
> 
> The picker can appear when the user has no intention of creating a new
> worktree, and the user cannot use the picker to create a new worktree
> inside another.
> 
> So, add two new explicit subcommands:
>     gui  - Run the gui if repository/worktree discovery succeeds, or die
>            with an error message, but never run the picker.
>     pick - First run the picker, regardless, then start the gui in
>            the chosen worktree.
> 
> Nothing in this changes the prior behavior, the alternates above must be
> explicitly selected to see any change.

OK.

> 
> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
> ---
>  git-gui.sh | 22 ++++++++++++++++++++--
>  1 file changed, 20 insertions(+), 2 deletions(-)
> 
> diff --git a/git-gui.sh b/git-gui.sh
> index 3a83dd5..c56aeef 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -1021,6 +1021,7 @@ proc load_config {include_global} {
>  ##
>  ## feature option selection
>  
> +set run_picker_on_error 1
>  if {[regexp {^git-(.+)$} [file tail $argv0] _junk subcommand]} {
>  	unset _junk
>  } else {
> @@ -1030,6 +1031,7 @@ if {$subcommand eq {gui.sh}} {
>  	set subcommand gui
>  }
>  if {$subcommand eq {gui} && [llength $argv] > 0} {
> +	set run_picker_on_error 0
>  	set subcommand [lindex $argv 0]
>  	set argv [lrange $argv 1 end]
>  }
> @@ -1047,6 +1049,7 @@ blame {
>  	disable_option multicommit
>  	disable_option branch
>  	disable_option transport
> +	set run_picker_on_error 0
>  }
>  citool {
>  	enable_option singlecommit
> @@ -1055,6 +1058,7 @@ citool {
>  	disable_option multicommit
>  	disable_option branch
>  	disable_option transport
> +	set run_picker_on_error 0
>  
>  	while {[llength $argv] > 0} {
>  		set a [lindex $argv 0]

Can we please use the available disable_option and enable_option feature
instead of a new variable. Just for consistency around repository discovery.

> @@ -1162,14 +1166,28 @@ proc pick_repo {} {
>  	set picked 1
>  }
>  
> +# run repository picker if explicitly requested
> +switch -- $subcommand {
> +	pick {
> +		pick_repo
> +		set subcommand gui
> +		set run_picker_on_error 0
> +	}
> +}
> +

It just feels wrong to have a new pick_repo call before repository
discovery. Can we not treat this case below as if regular repository
discovery failed and then end up in the existing call of pick_repo?

>  # find repository.
>  if {[catch {
>  	set _gitdir [git rev-parse --absolute-git-dir]
>  } err]} {
>  	if {[is_gitvars_error $err]} {
>  		exit 1
> -	} else {
> +	}
> +	if {$run_picker_on_error} {
>  		pick_repo
> +	} else {
> +		catch {wm withdraw .}
> +		error_popup [strcat [mc "Git directory not found:"] "\n\n$err"]
> +		exit 1
>  	}
>  }
>  
> @@ -3051,7 +3069,7 @@ gui {
>  	# fall through to setup UI for commits
>  }
>  default {
> -	set err "[mc usage:] $argv0 \[{blame|browser|citool}\]"
> +	set err "[mc usage:] $argv0 \[{blame|browser|citool|gui|pick}\]"
>  	if {[tk windowingsystem] eq "win32"} {
>  		wm withdraw .
>  		tk_messageBox -icon error -message $err \

We don't need to switch on the new subcommands?

As a follow-up to my comment on 04/11: How relevant is it that variable
$picked is set in a 'git gui pick' invocation?

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 11/11] git-gui: add gui and pick as explicit subcommands
  2026-05-16  8:18                     ` Johannes Sixt
@ 2026-05-16 15:42                       ` Mark Levedahl
  2026-05-19  8:21                         ` Johannes Sixt
  0 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-16 15:42 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: egg_mushroomcow, bootaina702, git



On 5/16/26 4:18 AM, Johannes Sixt wrote:
> Am 14.05.26 um 16:33 schrieb Mark Levedahl:
>> git-gui accepts subcommands blame | browser | citool, and assumes the
>> subcommand is 'gui' if none is actually given, But, git gui also has a
>> repository picker (choose_repository::pick) that can create a new
>> repository + worktree, or choose an existing one, switch to that, and
>> the run the gui. The user has no direct control over invoking the
>> picker, instead the picker is triggered by failure in the repository /
>> worktree discover process: this includes being started in a directory
>> not controlled by git, which is probably the intended use case.
>>
>> The picker can appear when the user has no intention of creating a new
>> worktree, and the user cannot use the picker to create a new worktree
>> inside another.
>>
>> So, add two new explicit subcommands:
>>     gui  - Run the gui if repository/worktree discovery succeeds, or die
>>            with an error message, but never run the picker.
>>     pick - First run the picker, regardless, then start the gui in
>>            the chosen worktree.
>>
>> Nothing in this changes the prior behavior, the alternates above must be
>> explicitly selected to see any change.
> OK.
>
>> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
>> ---
>>  git-gui.sh | 22 ++++++++++++++++++++--
>>  1 file changed, 20 insertions(+), 2 deletions(-)
>>
>> diff --git a/git-gui.sh b/git-gui.sh
>> index 3a83dd5..c56aeef 100755
>> --- a/git-gui.sh
>> +++ b/git-gui.sh
>> @@ -1021,6 +1021,7 @@ proc load_config {include_global} {
>>  ##
>>  ## feature option selection
>>  
>> +set run_picker_on_error 1
>>  if {[regexp {^git-(.+)$} [file tail $argv0] _junk subcommand]} {
>>  	unset _junk
>>  } else {
>> @@ -1030,6 +1031,7 @@ if {$subcommand eq {gui.sh}} {
>>  	set subcommand gui
>>  }
>>  if {$subcommand eq {gui} && [llength $argv] > 0} {
>> +	set run_picker_on_error 0
>>  	set subcommand [lindex $argv 0]
>>  	set argv [lrange $argv 1 end]
>>  }
>> @@ -1047,6 +1049,7 @@ blame {
>>  	disable_option multicommit
>>  	disable_option branch
>>  	disable_option transport
>> +	set run_picker_on_error 0
>>  }
>>  citool {
>>  	enable_option singlecommit
>> @@ -1055,6 +1058,7 @@ citool {
>>  	disable_option multicommit
>>  	disable_option branch
>>  	disable_option transport
>> +	set run_picker_on_error 0
>>  
>>  	while {[llength $argv] > 0} {
>>  		set a [lindex $argv 0]
> Can we please use the available disable_option and enable_option feature
> instead of a new variable. Just for consistency around repository discovery.
>
>> @@ -1162,14 +1166,28 @@ proc pick_repo {} {
>>  	set picked 1
>>  }
>>  
>> +# run repository picker if explicitly requested
>> +switch -- $subcommand {
>> +	pick {
>> +		pick_repo
>> +		set subcommand gui
>> +		set run_picker_on_error 0
>> +	}
>> +}
>> +
> It just feels wrong to have a new pick_repo call before repository
> discovery. Can we not treat this case below as if regular repository
> discovery failed and then end up in the existing call of pick_repo?

So, your suggestion is to create an error inside the catch clause, assure GIT_VAR and
GIT_WORK_TREE are unset so we don't throw and error message and abort, and then fall
through to the existing pick_repo clause?

I think this makes code less understandable and more complicated. As written, the user
selects the repo, and now the standard discovery runs except that the picker cannot be re-run.

I like the current structure better.

>>  # find repository.
>>  if {[catch {
>>  	set _gitdir [git rev-parse --absolute-git-dir]
>>  } err]} {
>>  	if {[is_gitvars_error $err]} {
>>  		exit 1
>> -	} else {
>> +	}
>> +	if {$run_picker_on_error} {
>>  		pick_repo
>> +	} else {
>> +		catch {wm withdraw .}
>> +		error_popup [strcat [mc "Git directory not found:"] "\n\n$err"]
>> +		exit 1
>>  	}
>>  }
>>  
>> @@ -3051,7 +3069,7 @@ gui {
>>  	# fall through to setup UI for commits
>>  }
>>  default {
>> -	set err "[mc usage:] $argv0 \[{blame|browser|citool}\]"
>> +	set err "[mc usage:] $argv0 \[{blame|browser|citool|gui|pick}\]"
>>  	if {[tk windowingsystem] eq "win32"} {
>>  		wm withdraw .
>>  		tk_messageBox -icon error -message $err \
> We don't need to switch on the new subcommands?
>
> As a follow-up to my comment on 04/11: How relevant is it that variable
> $picked is set in a 'git gui pick' invocation?
>
> -- Hannes
>
Repeating, I need to understand what "picked" really does in the gui. Will definitely
address your question.

Mark


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 11/11] git-gui: add gui and pick as explicit subcommands
  2026-05-16 15:42                       ` Mark Levedahl
@ 2026-05-19  8:21                         ` Johannes Sixt
  2026-05-19 18:45                           ` Mark Levedahl
  0 siblings, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-19  8:21 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 16.05.26 um 17:42 schrieb Mark Levedahl:
> On 5/16/26 4:18 AM, Johannes Sixt wrote:
>> Am 14.05.26 um 16:33 schrieb Mark Levedahl:
>>> diff --git a/git-gui.sh b/git-gui.sh
>>> index 3a83dd5..c56aeef 100755
>>> --- a/git-gui.sh
>>> +++ b/git-gui.sh
>>> @@ -1021,6 +1021,7 @@ proc load_config {include_global} {
>>>  ##
>>>  ## feature option selection
>>>  
>>> +set run_picker_on_error 1
>>>  if {[regexp {^git-(.+)$} [file tail $argv0] _junk subcommand]} {
>>>  	unset _junk
>>>  } else {
>>> @@ -1030,6 +1031,7 @@ if {$subcommand eq {gui.sh}} {
>>>  	set subcommand gui
>>>  }
>>>  if {$subcommand eq {gui} && [llength $argv] > 0} {
>>> +	set run_picker_on_error 0
>>>  	set subcommand [lindex $argv 0]
>>>  	set argv [lrange $argv 1 end]
>>>  }
>>> @@ -1047,6 +1049,7 @@ blame {
>>>  	disable_option multicommit
>>>  	disable_option branch
>>>  	disable_option transport
>>> +	set run_picker_on_error 0
>>>  }
>>>  citool {
>>>  	enable_option singlecommit
>>> @@ -1055,6 +1058,7 @@ citool {
>>>  	disable_option multicommit
>>>  	disable_option branch
>>>  	disable_option transport
>>> +	set run_picker_on_error 0
>>>  
>>>  	while {[llength $argv] > 0} {
>>>  		set a [lindex $argv 0]
>> Can we please use the available disable_option and enable_option feature
>> instead of a new variable. Just for consistency around repository discovery.
>>
>>> @@ -1162,14 +1166,28 @@ proc pick_repo {} {
>>>  	set picked 1
>>>  }
>>>  
>>> +# run repository picker if explicitly requested
>>> +switch -- $subcommand {
>>> +	pick {
>>> +		pick_repo
>>> +		set subcommand gui
>>> +		set run_picker_on_error 0
>>> +	}
>>> +}
>>> +
>> It just feels wrong to have a new pick_repo call before repository
>> discovery. Can we not treat this case below as if regular repository
>> discovery failed and then end up in the existing call of pick_repo?
> 
> So, your suggestion is to create an error inside the catch clause, assure GIT_VAR and
> GIT_WORK_TREE are unset so we don't throw and error message and abort, and then fall
> through to the existing pick_repo clause?


I think I would be happier with the structure

if not subcommand pick
    discover gitdir
    if error
        set subcommand pick

if subcommand pick
    pick_repo
    set subcommand gui

because this clarifies that pick_repo must erase all current traces of
GIT_DIR and GIT_WORK_TREE from the envionment and must complete with a
valid setup.

With the structure in the proposed patch

if subcommand pick
    pick_repo
    set subcommand gui

discover gitdir
if error
    pick_repo

we still need the same operation of pick_repo, but after it runs due to
a pick command, we go into "discover gitdir" mode in an already modified
environment, something that does not happen if pick_repo runs due to the
error in the gitdir discovery.

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 11/11] git-gui: add gui and pick as explicit subcommands
  2026-05-19  8:21                         ` Johannes Sixt
@ 2026-05-19 18:45                           ` Mark Levedahl
  2026-05-19 21:15                             ` Johannes Sixt
  0 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-19 18:45 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: egg_mushroomcow, bootaina702, git



On 5/19/26 4:21 AM, Johannes Sixt wrote:
> Am 16.05.26 um 17:42 
>
>
> I think I would be happier with the structure
>
> if not subcommand pick
>     discover gitdir
>     if error
>         set subcommand pick
>
> if subcommand pick
>     pick_repo
>     set subcommand gui
>
> because this clarifies that pick_repo must erase all current traces of
> GIT_DIR and GIT_WORK_TREE from the envionment and must complete with a
> valid setup.
>
> With the structure in the proposed patch
>
> if subcommand pick
>     pick_repo
>     set subcommand gui
>
> discover gitdir
> if error
>     pick_repo
>
> we still need the same operation of pick_repo, but after it runs due to
> a pick command, we go into "discover gitdir" mode in an already modified
> environment, something that does not happen if pick_repo runs due to the
> error in the gitdir discovery.
>
> -- Hannes
>

What I have now is

if (enabled gitdir discovery) {
    discover gitdir
    maybe an error occurs and gitdir remains {}
}

if (enabled pick && gitdir eq {}) {
    unset GIT_DIR .. (just to be friendly, could throw an error instead...)
    pick
    discover gitdir to VALIDATE pick gave us a good thing
}

if no gitdir {
    error No Repository
}

then on to worktree discovery (which also validates what pick returns as pick may not have
done so).

So, we can independently enable normal discovery or pick, and with both enabled pick can
be used to recover from an error in normal discovery. Either way, there is only one block
of code running pick, it is not a separate proc invoked multiple places. I hope this
scratches your itch.

Still scrubbing things, should send out in a day or two.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 11/11] git-gui: add gui and pick as explicit subcommands
  2026-05-19 18:45                           ` Mark Levedahl
@ 2026-05-19 21:15                             ` Johannes Sixt
  0 siblings, 0 replies; 134+ messages in thread
From: Johannes Sixt @ 2026-05-19 21:15 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 19.05.26 um 20:45 schrieb Mark Levedahl:
> What I have now is
> 
> if (enabled gitdir discovery) {
>     discover gitdir
>     maybe an error occurs and gitdir remains {}
> }
> 
> if (enabled pick && gitdir eq {}) {
>     unset GIT_DIR .. (just to be friendly, could throw an error instead...)
>     pick
>     discover gitdir to VALIDATE pick gave us a good thing
> }
> 
> if no gitdir {
>     error No Repository
> }
> 
> then on to worktree discovery (which also validates what pick returns as pick may not have
> done so).

Sounds good.

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v1 00/11] Improve git gui operation without a worktree
  2026-05-14 14:33                 ` [PATCH v1 00/11] Improve git gui operation without a worktree Mark Levedahl
                                     ` (10 preceding siblings ...)
  2026-05-14 14:33                   ` [PATCH v1 11/11] git-gui: add gui and pick as explicit subcommands Mark Levedahl
@ 2026-05-16  8:28                   ` Johannes Sixt
  2026-05-20 20:23                   ` [PATCH v2 " Mark Levedahl
  12 siblings, 0 replies; 134+ messages in thread
From: Johannes Sixt @ 2026-05-16  8:28 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 14.05.26 um 16:33 schrieb Mark Levedahl:
> git gui has a number of inter-related problems that result in problems
> during startup from anything but a checked out worktree pointing at a
> valid git repository. Some of the symptoms are:
> - blame / browser subcommands, and launching gitk, are intended to be
>   useful without a worktree, but fail to work.
> - unlike git, git-gui is supposed to use the parent directory as a
>   worktree if started from the .git subdirectory in the very common
>   single worktree + embedded git repository format. This does not
>   work.
> - git-gui includes a repository picker allowing a user select a worktree
>   from a list and/or start a new repo+worktree: this dialog appears at
>   unexpected times, masking useful error feedback on configuration
>   problems.
> 
> This patch series addresses the above issues, substantially rewriting
> the blame / browser command line process, the initial repository and
> worktree discovery processes, and using git rev-parse when possible to
> handle repository / worktree discovery including any specification of
> GIT_DIR or GIT_WORK_TREE to reduce the future likelihood of conflict
> with command line git. This also allows explicit user control to avoid
> the repository picker masking a configuration error.

OK. Overall, this goes in the right direction. There are still open
questions and potential problems with this implementation. We also
disagree in a few details; see my comments on the patches.

> 
> Note: I question why git-gui ever exports GIT_WORK_TREE. If it is not
> empty, that is the current directory when startup is complete and any
> git command will use the current directory as the worktree.

I fully agree with this.

> If empty,
> there is no worktree and the current directory should be (and after this
> series, is) at the toplevel of the gitdir: again, there is nothing to
> communicate to another process.

Here I disagree. We should not need to change directory if no working
tree was found.

> If a process being launched needs a
> different worktree, that should be the startup directory given to the
> process without changing git-gui's current directory.

I haven't thought this through, but this sounds very reasonable.

> 
> Mark Levedahl (11):
>   git-gui: allow specifying path '.' to the browser
>   git-gui: refactor browser / blame argument parsing
>   git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE
>   git-gui: put choose_repository::pick in a proc
>   git-gui: use --absolute-git-dir
>   git gui: GIT_DIR / GIT_WORK_TREE make any discovery error fatal
>   git-gui: use rev-parse exclusively to find a repository
>   git-gui: simplify [is_bare] to report if a worktree is known
>   git-gui: support using repository parent dir as a worktree
>   git-gui: improve worktree discovery
>   git-gui: add gui and pick as explicit subcommands
> 
>  git-gui.sh | 276 ++++++++++++++++++++++++++---------------------------
>  1 file changed, 135 insertions(+), 141 deletions(-)
> 

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v2 00/11] Improve git gui operation without a worktree
  2026-05-14 14:33                 ` [PATCH v1 00/11] Improve git gui operation without a worktree Mark Levedahl
                                     ` (11 preceding siblings ...)
  2026-05-16  8:28                   ` [PATCH v1 00/11] Improve git gui operation without a worktree Johannes Sixt
@ 2026-05-20 20:23                   ` Mark Levedahl
  2026-05-20 20:24                     ` [PATCH v2 01/11] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE Mark Levedahl
                                       ` (12 more replies)
  12 siblings, 13 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-20 20:23 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

git gui has a number of inter-related problems that result in problems
during startup from anything but a checked out worktree pointing at a
valid git repository. Some of the symptoms are:
- blame / browser subcommands, and launching gitk, are intended to be
  useful without a worktree, but fail to work.
- unlike git, git-gui is supposed to use the parent directory as a
  worktree if started from the .git subdirectory in the very common
  single worktree + embedded git repository format. This does not
  work.
- git-gui includes a repository picker allowing a user to select a
  worktree from a list and/or start a new repo+worktree: this dialog can
  appear at unexpected times, masking useful error feedback on
  configuration problems.

This patch series addresses the above issues, substantially rewriting
the initial repository/worktree process to rely upon git rev-parse so
that git's knowledge of access rules, repository configuration, and use
of GIT_DIR / GIT_WORK_TREE (or git --gitdir / --work-tree) is used
throughout, replacing code largely based upon what git did in 2008. This
also means that git gui will naturally gain any new rules implmented in
git-core.

With this, git-gui only exports GIT_WORK_TREE when non-empty.
GIT_WORK_TREE is needed, and must be exported, if the user is overriding
core.worktree in the git repository. But, GIT_WORK_TREE cannot be used
to specify the lack of a worktree, so exporting an empty GIT_WORK_TREE
is one of the problems fixed by this series.

v2 of this series is a very substantial rewrite driven by j6t's review,
with patches reoranized and squashed, interfaces to the repository
chooser changed, a different code structure to allow user control of the
repository picker, a different approach to fixing the command line
parser for blame / browser, and other more minor changes. Patches
for fixing blame / browser are now after all discovery refactoring as
they cannot be tested without some of those fixes.

Many subtle things are fixed beyond the list at the top, including
better compatibility with git blame and repeatable browser / blame
operation for specific revs not in the worktree, regardless of the
worktree state. j6t indicated that in the git-gui project, the following
fails in the current release:

cd lib
GIT_DIR=$PWD/../.git GIT_WORK_TREE=$PWD/.. ../git-gui.sh browser origin/master .

This is due to a _prefix issue, and is fixed as of the patch
     git-gui: use git rev-parse for worktree discovery

Mark Levedahl (11):
  git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE
  git-gui: return status from choose_repository::pick
  git-gui: use --absolute-git-dir
  git-gui: use rev-parse exclusively to find a repository
  git-gui: simplify [is_bare] to report if a worktree is known
  git-gui: use git rev-parse for worktree discovery
  git-gui: try harder to find worktree from gitdir
  git-gui: use HEAD as current branch when detached (bug fix)
  git-gui: allow specifying path '.' to the browser
  git-gui: adapt blame/browser parsing for bare operation
  git-gui: add gui and pick as explicit subcommands

 git-gui.sh                | 412 +++++++++++++++++++++++---------------
 lib/choose_repository.tcl |  21 +-
 2 files changed, 257 insertions(+), 176 deletions(-)

Interdiff against v1:
diff --git a/git-gui.sh b/git-gui.sh
index c56aeeff88..299c1a0292 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -648,6 +648,9 @@ proc load_current_branch {} {
 
 	set current_branch [git branch --show-current]
 	set is_detached [expr [string length $current_branch] == 0]
+	if {$is_detached} {
+		set current_branch {HEAD}
+	}
 }
 
 auto_load tk_optionMenu
@@ -1021,7 +1024,8 @@ proc load_config {include_global} {
 ##
 ## feature option selection
 
-set run_picker_on_error 1
+enable_option picker
+enable_option gitdir_discovery
 if {[regexp {^git-(.+)$} [file tail $argv0] _junk subcommand]} {
 	unset _junk
 } else {
@@ -1031,9 +1035,11 @@ if {$subcommand eq {gui.sh}} {
 	set subcommand gui
 }
 if {$subcommand eq {gui} && [llength $argv] > 0} {
-	set run_picker_on_error 0
 	set subcommand [lindex $argv 0]
 	set argv [lrange $argv 1 end]
+	if {$subcommand eq {gui}} {
+		disable_option picker
+	}
 }
 
 enable_option multicommit
@@ -1049,7 +1055,7 @@ blame {
 	disable_option multicommit
 	disable_option branch
 	disable_option transport
-	set run_picker_on_error 0
+	disable_option picker
 }
 citool {
 	enable_option singlecommit
@@ -1058,7 +1064,7 @@ citool {
 	disable_option multicommit
 	disable_option branch
 	disable_option transport
-	set run_picker_on_error 0
+	disable_option picker
 
 	while {[llength $argv] > 0} {
 		set a [lindex $argv 0]
@@ -1081,6 +1087,9 @@ citool {
 		set argv [lrange $argv 1 end]
 	}
 }
+pick {
+	disable_option gitdir_discovery
+}
 }
 
 ######################################################################
@@ -1104,21 +1113,39 @@ unset argv0dir
 ##
 ## repository setup
 
-proc is_parent_worktree {} {
-	# Directory 'parent' of a repository named 'parent/.git' might be the worktree
-	set ok 0
+proc find_worktree_from_gitdir {} {
+	# Directory 'parent' of a repository named 'parent/.git' might be the worktree.
+	# Assure parent is a worktree and using the git repository already discovered.
+	# Also, handle case of being in a worktree's gitdir, where file "gitdir" points to
+	# gitlink file .git in the real worktree.
+	set worktree {}
 	if {[file tail $::_gitdir] eq {.git}} {
-		set gitdir_parent [file join $::_gitdir {..}]
-		set expected_worktree [file normalize $gitdir_parent]
-		catch {set git_worktree [git -C $gitdir_parent rev-parse --show-toplevel]}
-		if {[string compare $expected_worktree $git_worktree] == 0} {
-			set ::_prefix {}
-			set ::_gitworktree $git_worktree
-			cd $git_worktree
-			set ok 1
+		if {[catch {
+			set gitdir_parent [file dirname $::_gitdir]
+			set worktree [git -C $gitdir_parent rev-parse --show-toplevel]
+			set parent_gitdir [git -C $worktree rev-parse --absolute-git-dir]
+			if {$::_gitdir ne $parent_gitdir} {
+				set worktree {}
+			}
+		}]} {
+			set worktree {}
+		}
+	} elseif [file exists {gitdir}] {
+		if {[catch {
+			set fd_gitdir [open {gitdir} {r}]
+			set gitlink_parent [file dirname [read $fd_gitdir]]
+			catch {close $fd_gitdir}
+			set worktree [git -C $gitlink_parent rev-parse --show-toplevel]
+			set parent_gitdir [git -C $worktree rev-parse --absolute-git-dir]
+			if {$::_gitdir ne $parent_gitdir} {
+				set worktree {}
+			}
+		}]} {
+			catch {close $fd_gitdir}
+			set worktree {}
 		}
 	}
-	return $ok
+	return $worktree
 }
 
 proc is_gitvars_error {err} {
@@ -1155,62 +1182,76 @@ proc unset_gitdir_vars {} {
 	catch {unset env(GIT_WORK_TREE)}
 }
 
+# find repository.
+set _gitdir {}
+if {[is_enabled gitdir_discovery]} {
+	if {[catch {
+		set _gitdir [git rev-parse --absolute-git-dir]
+	} err]} {
+		if {[is_gitvars_error $err]} {
+			exit 1
+		}
+		set _gitdir {}
+	}
+}
+
 set picked 0
-proc pick_repo {} {
+if {$_gitdir eq {} && [is_enabled picker]} {
 	unset_gitdir_vars
 	load_config 1
 	apply_config
-	choose_repository::pick
-	set _gitdir [git rev-parse --absolute-git-dir]
-	set _prefix {}
+	if {![choose_repository::pick]} {
+		exit 1
+	}
+	if {[catch {
+		set _gitdir [git rev-parse --absolute-git-dir]
+	} err]} {
+		catch {wm withdraw .}
+		error_popup [strcat [mc "Unusable repo/worktree:"] " [pwd] "\n\n$err"]
+	}
 	set picked 1
 }
 
-# run repository picker if explicitly requested
-switch -- $subcommand {
-	pick {
-		pick_repo
-		set subcommand gui
-		set run_picker_on_error 0
-	}
-}
-
-# find repository.
-if {[catch {
-	set _gitdir [git rev-parse --absolute-git-dir]
-} err]} {
-	if {[is_gitvars_error $err]} {
-		exit 1
-	}
-	if {$run_picker_on_error} {
-		pick_repo
-	} else {
-		catch {wm withdraw .}
-		error_popup [strcat [mc "Git directory not found:"] "\n\n$err"]
-		exit 1
-	}
+if {$_gitdir eq {}} {
+	catch {wm withdraw .}
+	error_popup [strcat [mc "Git directory not found:"] "\n\n$err"]
+	exit 1
 }
 
 # find worktree, continue without if not required
 if {[catch {
 	set _gitworktree [git rev-parse --show-toplevel]
 	set _prefix [git rev-parse --show-prefix]
-	cd $_gitworktree
 } err]} {
 	if {[is_gitvars_error $err]} {
 		exit 1
 	}
 	set _gitworktree {}
 	set _prefix {}
-	if {[is_enabled bare]} {
-		cd $_gitdir
-	} elseif {![is_parent_worktree]} {
-		catch {wm withdraw .}
-		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n" $_gitdir]
-		exit 1
+}
+
+if {[is_bare]} {
+	# Maybe we are in an embedded or worktree specific gitdir
+	if {[set _gitworktree [find_worktree_from_gitdir]] ne {}} {
+		set _prefix {}
 	}
 }
 
+if {![is_bare]} {
+	if {[catch {
+		cd $_gitworktree
+	} err]} {
+		catch {wm withdraw .}
+		error_popup [strcat [mc "Cannot change to discovered worktree: "] \
+			"$_gitworktree" "\n\n$err"]
+		exit 1;
+	}
+} elseif {![is_enabled bare]} {
+	catch {wm withdraw .}
+	error_popup [strcat [mc "Cannot use bare repository:"] "\n\n" $_gitdir]
+	exit 1
+}
+
 # repository and worktree config are complete, export them
 set_gitdir_vars
 
@@ -1229,8 +1270,6 @@ if {$hashalgorithm eq "sha1"} {
 load_config 0
 apply_config
 
-
-# Derive a human-readable repository name
 set _reponame [file split [file normalize $_gitdir]]
 if {[lindex $_reponame end] eq {.git}} {
 	set _reponame [lindex $_reponame end-1]
@@ -2035,7 +2074,7 @@ proc incr_font_size {font {amt 1}} {
 
 proc do_gitk {revs {is_submodule false}} {
 	global current_diff_path file_states current_diff_side ui_index
-	global _gitdir _gitworktree
+	global _gitworktree
 
 	# -- Always start gitk through whatever we were loaded with.  This
 	#    lets us bypass using shell process on Windows systems.
@@ -2045,8 +2084,6 @@ proc do_gitk {revs {is_submodule false}} {
 	if {$exe eq {}} {
 		error_popup [mc "Couldn't find gitk in PATH"]
 	} else {
-		global env
-
 		set pwd [pwd]
 
 		if {!$is_submodule} {
@@ -2105,9 +2142,6 @@ proc do_git_gui {} {
 	if {$exe eq {}} {
 		error_popup [mc "Couldn't find git gui in PATH"]
 	} else {
-		global env
-		global _gitdir _gitworktree
-
 		# see note in do_gitk about unsetting these vars when
 		# running tools in a submodule
 		unset_gitdir_vars
@@ -2992,77 +3026,135 @@ proc normalize_relpath {path} {
 	if {$elements ne {}} {
 		return [eval file join $elements]
 	} else {
-		return {}
+		return {./}
 	}
 }
 
+proc find_path_type {head path} {
+	if {$path eq {./}} {
+		# the root-tree exists in every rev, ls-tree gives data on the contents,
+		# not the type of tree itself. So, if the rev exists, return {tree}
+		if {[catch {set objtype [git ls-tree $head]}]} {
+			set objtype {}
+		} else {
+			set objtype {tree}
+		}
+	} else {
+		# test that the path exists in head, ls-tree gives info on the path only
+		if {[catch {set objtype [git ls-tree {--format=%(objecttype)} $head $path]}]} {
+			set objtype {}
+		}
+	}
+	return $objtype
+}
+
 # -- Not a normal commit type invocation?  Do that instead!
 #
 switch -- $subcommand {
 browser -
 blame {
 	if {$subcommand eq "blame"} {
-		set subcommand_args {[--line=<num>] rev? path}
+		set subcommand_args {[--line=<num>] <[rev] [--] filename | [--] filename rev>}
+		set required_objtype blob
 	} else {
-		set subcommand_args {rev? path}
+		set subcommand_args {<[rev] [--] directory | [--] directory rev>}
+		set required_objtype tree
 	}
-	if {$argv eq {}} usage
-	set head {}
-	set path {}
-	set jump_spec {}
+	set maxargs [llength $subcommand_args]
 	set nargs [llength $argv]
-	if {$nargs < 1} {
-		usage
-	}
-	set argn 0
-	foreach a $argv {
-		set argn [expr {$argn + 1}]
+	if {$nargs < 1 || $nargs > $maxargs} usage
+	set head {}
+	set althead {}
+	set path {}
+	set altpath {}
+	set canswap 1
+	set jump_spec {}
 
-		if {$argn < $nargs} {
-			# revision or line number
-			if {[regexp {^--line=(\d+)$} $a a lnum]} {
-				set jump_spec [list $lnum]
+	# assume: [--line=num] [head] [--] path as the possible arguments, in order.
+	# head and path may need a swap later.
+	for {set iarg 0} {$iarg < $nargs} {incr iarg} {
+		set arg [lindex $argv $iarg]
+		if {$arg eq {--}} {
+			# next arg is the path, prevent or FORCE swap?
+			if {$iarg == $nargs - 2} {
+				set canswap 0
+			} elseif {$iarg == $nargs - 3} {
+				set canswap 2
 			} else {
-				set head $a
+				usage
 			}
+		} elseif {[regexp {^--line=(\d+)$} $arg arg lnum]} {
+			# --line can only be the first arg
+			if {$iarg != 0 || $maxargs < 4} usage
+			set jump_spec [list $lnum]
+		} elseif {$iarg == $nargs - 1} {
+			# assume final argument is path
+			set path [normalize_relpath [file join $_prefix $arg]]
+			set althead $arg
+		} elseif {$head eq {}} {
+			# assume the other argument is head
+			set head $arg
+			set altpath [normalize_relpath [file join $_prefix $arg]]
 		} else {
-			set path [normalize_relpath $a]
+			usage
 		}
 	}
 
+	# no swapping allowed if head not given, use current branch (HEAD)
 	if {$head eq {}} {
 		load_current_branch
 		set head $current_branch
-	} else {
-		if {[regexp [string map "@@ [expr $hashlength - 1]" {^[0-9a-f]{1,@@}$}] $head]} {
-			if {[catch {
-				set head [git rev-parse --verify $head]
-			} err]} {
-				if {[tk windowingsystem] eq "win32"} {
-					tk_messageBox -icon error -title [mc Error] -message $err
-				} else {
-					puts stderr $err
-				}
-				exit 1
-			}
+		set canswap 0
+	}
+
+	# -- before "rev" arg means we got -- path head
+	if {$canswap == 2} {
+		set head $althead
+		set path $altpath
+		set canswap 0
+	}
+
+	set objtype [find_path_type $head $path]
+	if {$objtype eq {} && $canswap} {
+		set objtype [find_path_type $althead $altpath]
+		if {$objtype ne {}} {
+			set head $althead
+			set path $altpath
 		}
-		set current_branch $head
+	}
+	set current_branch $head
+
+	# check that path exists in head, and objtype matches need
+	if {$objtype ne $required_objtype} {
+		switch -- $required_objtype {
+			tree {set err [strcat \
+				[mc "'%s' is not a directory in rev '%s'" $path $head]]}
+			blob {set err [strcat \
+				[mc "'%s' is not a filename in rev '%s'" $path $head]]}
+		}
+		if {[tk windowingsystem] eq "win32"} {
+			catch {wm withdraw .}
+			error_popup $err
+		} else {
+			puts stderr $err
+		}
+		exit 1
 	}
 
 	wm deiconify .
 	switch -- $subcommand {
 	browser {
-		if {$jump_spec ne {}} usage
 		browser::new $head $path
 	}
-	blame   {
+	blame {
 		blame::new $head $path $jump_spec
 	}
 	}
 	return
 }
 citool -
-gui {
+gui -
+pick {
 	if {[llength $argv] != 0} {
 		usage
 	}
diff --git a/lib/choose_repository.tcl b/lib/choose_repository.tcl
index 7e1462a20c..4b06afee93 100644
--- a/lib/choose_repository.tcl
+++ b/lib/choose_repository.tcl
@@ -15,7 +15,7 @@ field w_recentlist ; # Listbox containing recent repositories
 field w_localpath  ; # Entry widget bound to local_path
 
 field done              0 ; # Finished picking the repository?
-field clone_ok      false ; # clone succeeeded
+field pick_ok           0 ; # true if repo pick/clone succeeded
 field local_path       {} ; # Where this repository is locally
 field origin_url       {} ; # Where we are cloning from
 field origin_name  origin ; # What we shall call 'origin'
@@ -220,6 +220,8 @@ constructor pick {} {
 	if {$top eq {.}} {
 		eval destroy [winfo children $top]
 	}
+
+	return $pick_ok
 }
 
 method _center {} {
@@ -327,8 +329,7 @@ method _git_init {} {
 	}
 
 	_append_recentrepos [pwd]
-	set ::_gitdir .git
-	set ::_prefix {}
+	set pick_ok 1
 	return 1
 }
 
@@ -409,6 +410,7 @@ method _do_new2 {} {
 	if {![_git_init $this]} {
 		return
 	}
+	set pick_ok 1
 	set done 1
 }
 
@@ -621,7 +623,7 @@ method _do_clone2 {} {
 	}
 
 	tkwait variable @done
-	if {!$clone_ok} {
+	if {!$pick_ok} {
 		error_popup [mc "Clone failed."]
 		return
 	}
@@ -632,18 +634,12 @@ method _do_clone2_done {ok} {
 	if {$ok} {
 		if {[catch {
 			cd $local_path
-			set ::_gitdir .git
-			set ::_prefix {}
 			_append_recentrepos [pwd]
 		} err]} {
 			set ok 0
 		}
 	}
-	if {!$ok} {
-		set ::_gitdir {}
-		set ::_prefix {}
-	}
-	set clone_ok $ok
+	set pick_ok $ok
 	set done 1
 }
 
@@ -721,8 +717,7 @@ method _do_open2 {} {
 	}
 
 	_append_recentrepos [pwd]
-	set ::_gitdir $actualgit
-	set ::_prefix {}
+	set pick_ok 1
 	set done 1
 }
 
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* [PATCH v2 01/11] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE
  2026-05-20 20:23                   ` [PATCH v2 " Mark Levedahl
@ 2026-05-20 20:24                     ` Mark Levedahl
  2026-05-22  8:06                       ` Johannes Sixt
  2026-05-20 20:24                     ` [PATCH v2 02/11] git-gui: return status from choose_repository::pick Mark Levedahl
                                       ` (11 subsequent siblings)
  12 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-20 20:24 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

git-gui unconditionally exports _gitdir as GIT_DIR, and _gitworktree as
GIT_WORK_TREE, to the environment, and furthermore unconditionally
unsets these environment variables in many

git gui must have a repository, so _gitdir can never be empty and its
export is always valid if repository discovery completes successfully.

git gui might not find a worktree, so _gitworktree can be empty. While
having no worktree is valid for blame/browser subcommands, exporting
GIT_WORK_TREE=<empty> is not valid. Rather, an empty GIT_WORK_TREE
raises errors in git builtins, for instance 'git branch --show-current'
as used by git, and causes breakage. This is one cause of git blame /
git browser not working without a worktree.

A user may set GIT_DIR and/or GIT_WORK_TREE to override git's normal
discovery rules, including repository configuration of core.worktree
and/or worktree specific gitdirs. It is always safe to export the
absolute pathnames of the discovered values, even though they may not be
needed. However, the gitdir might not be found from the worktree without
GIT_DIR being set. Furthermore, the worktree defined by the discovered
gitdir might be overridden by GIT_WORK_TREE set before git-gui started.
So, it is also sometimes necessary that one or both of these variables
is set.

So, let's provide two procs, one to unset GIT_DIR / GIT_WORK_TREE if
they are set, one to set GIT_DIR and, if not empty, GIT_WORK_TREE,  so
all call sites do the same thing, and problems with _gitworktree == {}
are avoided.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 39 +++++++++++++++++++++++----------------
 1 file changed, 23 insertions(+), 16 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 23fe76e498..4ba25da7b6 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1122,6 +1122,22 @@ unset argv0dir
 ##
 ## repository setup
 
+proc set_gitdir_vars {} {
+	global _gitdir _gitworktree env
+	if {$_gitdir ne {}} {
+		set env(GIT_DIR) $_gitdir
+	}
+	if {$_gitworktree ne {}} {
+		set env(GIT_WORK_TREE) $_gitworktree
+	}
+}
+
+proc unset_gitdir_vars {} {
+	global env
+	catch {unset env(GIT_DIR)}
+	catch {unset env(GIT_WORK_TREE)}
+}
+
 set picked 0
 if {[catch {
 		set _gitdir $env(GIT_DIR)
@@ -1207,8 +1223,8 @@ if {[lindex $_reponame end] eq {.git}} {
 	set _reponame [lindex $_reponame end]
 }
 
-set env(GIT_DIR) $_gitdir
-set env(GIT_WORK_TREE) $_gitworktree
+# Export the final paths
+set_gitdir_vars
 
 ######################################################################
 ##
@@ -2007,7 +2023,7 @@ proc incr_font_size {font {amt 1}} {
 
 proc do_gitk {revs {is_submodule false}} {
 	global current_diff_path file_states current_diff_side ui_index
-	global _gitdir _gitworktree
+	global _gitworktree
 
 	# -- Always start gitk through whatever we were loaded with.  This
 	#    lets us bypass using shell process on Windows systems.
@@ -2017,8 +2033,6 @@ proc do_gitk {revs {is_submodule false}} {
 	if {$exe eq {}} {
 		error_popup [mc "Couldn't find gitk in PATH"]
 	} else {
-		global env
-
 		set pwd [pwd]
 
 		if {!$is_submodule} {
@@ -2050,13 +2064,11 @@ proc do_gitk {revs {is_submodule false}} {
 			# TODO we could make life easier (start up faster?) for gitk
 			# by setting these to the appropriate values to allow gitk
 			# to skip the heuristics to find their proper value
-			unset env(GIT_DIR)
-			unset env(GIT_WORK_TREE)
+			unset_gitdir_vars
 		}
 		safe_exec_bg [concat $cmd $revs "--" "--"]
 
-		set env(GIT_DIR) $_gitdir
-		set env(GIT_WORK_TREE) $_gitworktree
+		set_gitdir_vars
 		cd $pwd
 
 		if {[info exists main_status]} {
@@ -2079,21 +2091,16 @@ proc do_git_gui {} {
 	if {$exe eq {}} {
 		error_popup [mc "Couldn't find git gui in PATH"]
 	} else {
-		global env
-		global _gitdir _gitworktree
-
 		# see note in do_gitk about unsetting these vars when
 		# running tools in a submodule
-		unset env(GIT_DIR)
-		unset env(GIT_WORK_TREE)
+		unset_gitdir_vars
 
 		set pwd [pwd]
 		cd $current_diff_path
 
 		safe_exec_bg [concat $exe gui]
 
-		set env(GIT_DIR) $_gitdir
-		set env(GIT_WORK_TREE) $_gitworktree
+		set_gitdir_vars
 		cd $pwd
 
 		set status_operation [$::main_status \
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 01/11] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE
  2026-05-20 20:24                     ` [PATCH v2 01/11] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE Mark Levedahl
@ 2026-05-22  8:06                       ` Johannes Sixt
  2026-05-22 11:54                         ` Mark Levedahl
  0 siblings, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-22  8:06 UTC (permalink / raw)
  To: Mark Levedahl, git; +Cc: egg_mushroomcow, bootaina702

Am 20.05.26 um 22:24 schrieb Mark Levedahl:
> git-gui unconditionally exports _gitdir as GIT_DIR, and _gitworktree as
> GIT_WORK_TREE, to the environment, and furthermore unconditionally
> unsets these environment variables in many

"many cases"?

> 
> git gui must have a repository, so _gitdir can never be empty and its
> export is always valid if repository discovery completes successfully.

_gitdir cannot be empty, so we should be able to drop the conditionals
around 'set env(GIT_DIR) $_gitdir'.

> 
> git gui might not find a worktree, so _gitworktree can be empty. While
> having no worktree is valid for blame/browser subcommands, exporting
> GIT_WORK_TREE=<empty> is not valid. Rather, an empty GIT_WORK_TREE
> raises errors in git builtins, for instance 'git branch --show-current'
> as used by git, and causes breakage. This is one cause of git blame /
> git browser not working without a worktree.
> 
> A user may set GIT_DIR and/or GIT_WORK_TREE to override git's normal
> discovery rules, including repository configuration of core.worktree
> and/or worktree specific gitdirs. It is always safe to export the
> absolute pathnames of the discovered values, even though they may not be
> needed. However, the gitdir might not be found from the worktree without
> GIT_DIR being set. Furthermore, the worktree defined by the discovered
> gitdir might be overridden by GIT_WORK_TREE set before git-gui started.
> So, it is also sometimes necessary that one or both of these variables
> is set.

While all you say here is true, the actual reason for the dance is more
like the simpler: provide a clean slate for the new process and return
to the old state after it has been started.

> 
> So, let's provide two procs, one to unset GIT_DIR / GIT_WORK_TREE if
> they are set, one to set GIT_DIR and, if not empty, GIT_WORK_TREE,  so
> all call sites do the same thing, and problems with _gitworktree == {}
> are avoided.

That being said, I propose the two patches below (pasted here for
review), after which we do not need these functions anymore IMHO
because the call sites are one-liners around GIT_DIR anyway.

The commits are available here:

git fetch https://github.com/j6t/git-gui.git js/unset-git-work-tree
https://github.com/j6t/git-gui/tree/js/unset-git-work-tree

------ 8< ------
Subject: [PATCH 1/2] git-gui: remove unnecessary 'cd $_gitworktree' from do_gitk

In the procedure that invokes Gitk, we have a 'cd $_gitworktree'. Such
a change of the current directory is not necessary, because

- if we have a working tree, then the startup routine has already
  changed the current directory to the root of the working tree, which
  *is* $_gitworktree; or

- if we are in a bare repository, then there is no point in changing
  the current directory anywhere. (And $_gitworktree is empty.)

Signed-off-by: Johannes Sixt <j6t@kdbg•org>
---
 git-gui.sh | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 23fe76e498bd..8d2b02b13fa0 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -2021,11 +2021,7 @@ proc do_gitk {revs {is_submodule false}} {
 
 		set pwd [pwd]
 
-		if {!$is_submodule} {
-			if {![is_bare]} {
-				cd $_gitworktree
-			}
-		} else {
+		if {$is_submodule} {
 			cd $current_diff_path
 			if {$revs eq {--}} {
 				set s $file_states($current_diff_path)
-- 
2.54.0.215.g4fe990ec16

------ 8< ------
Subject: [PATCH 2/2] git-gui: operate git commands without GIT_WORK_TREE

The manual page of the git command states about the --git-dir option:

   Specifying the location of the ".git" directory using this option
   (or GIT_DIR environment variable) turns off the repository
   discovery [...], and tells Git that you are at the top level of
   the working tree.

Use this to our advantage:

- Set GIT_DIR in the environment to the value that was discovered, so
  that the invoked git commands operate on the same repository
  database that Git GUI uses even after it changes the working
  directory.

- After changing the working directory to the top level of the working
  tree, ensure that GIT_WORK_TREE is not set, because, as per
  documentation, all git invocations from then on will assume that the
  current working directory is also the top level working tree.

- Remove the now obsolete GIT_WORK_TREE dance when subordinate Gitk or
  Git GUI are invoked for a submodule.

Do keep the state of GIT_WORK_TREE if we are in a bare repository,
because Git GUI is not interested in the worktree at all, as no commit
mode is possible in a bare repository.

This avoids cases where an empty GIT_WORK_TREE was exported into the
environment, most notably by a call of `git gui blame HEAD file` in a
bare repository. (Although, this particular error is currently masked
by an earlier failure in `rev-parse --show-toplevel`, which requires a
working tree.)

Signed-off-by: Johannes Sixt <j6t@kdbg•org>
---
 git-gui.sh | 17 +++++++----------
 1 file changed, 7 insertions(+), 10 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 8d2b02b13fa0..3819f8be2211 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1183,6 +1183,7 @@ if {$_prefix ne {}} {
 		exit 1
 	}
 	set _gitworktree [pwd]
+	catch {unset env(GIT_WORK_TREE)}
 	unset cdup
 } elseif {![is_enabled bare]} {
 	if {[is_bare]} {
@@ -1199,6 +1200,7 @@ if {$_prefix ne {}} {
 		exit 1
 	}
 	set _gitworktree [pwd]
+	catch {unset env(GIT_WORK_TREE)}
 }
 set _reponame [file split [file normalize $_gitdir]]
 if {[lindex $_reponame end] eq {.git}} {
@@ -1208,7 +1210,6 @@ if {[lindex $_reponame end] eq {.git}} {
 }
 
 set env(GIT_DIR) $_gitdir
-set env(GIT_WORK_TREE) $_gitworktree
 
 ######################################################################
 ##
@@ -2007,7 +2008,7 @@ proc incr_font_size {font {amt 1}} {
 
 proc do_gitk {revs {is_submodule false}} {
 	global current_diff_path file_states current_diff_side ui_index
-	global _gitdir _gitworktree
+	global _gitdir
 
 	# -- Always start gitk through whatever we were loaded with.  This
 	#    lets us bypass using shell process on Windows systems.
@@ -2041,18 +2042,16 @@ proc do_gitk {revs {is_submodule false}} {
 				}
 				set revs $old_sha1...$new_sha1
 			}
-			# GIT_DIR and GIT_WORK_TREE for the submodule are not the ones
-			# we've been using for the main repository, so unset them.
+			# GIT_DIR for the submodule is not the one we've been using for
+			# the main repository, so unset it. (GIT_WORK_TREE is already unset.)
 			# TODO we could make life easier (start up faster?) for gitk
 			# by setting these to the appropriate values to allow gitk
 			# to skip the heuristics to find their proper value
 			unset env(GIT_DIR)
-			unset env(GIT_WORK_TREE)
 		}
 		safe_exec_bg [concat $cmd $revs "--" "--"]
 
 		set env(GIT_DIR) $_gitdir
-		set env(GIT_WORK_TREE) $_gitworktree
 		cd $pwd
 
 		if {[info exists main_status]} {
@@ -2076,12 +2075,11 @@ proc do_git_gui {} {
 		error_popup [mc "Couldn't find git gui in PATH"]
 	} else {
 		global env
-		global _gitdir _gitworktree
+		global _gitdir
 
-		# see note in do_gitk about unsetting these vars when
+		# see note in do_gitk about unsetting this variable when
 		# running tools in a submodule
 		unset env(GIT_DIR)
-		unset env(GIT_WORK_TREE)
 
 		set pwd [pwd]
 		cd $current_diff_path
@@ -2089,7 +2087,6 @@ proc do_git_gui {} {
 		safe_exec_bg [concat $exe gui]
 
 		set env(GIT_DIR) $_gitdir
-		set env(GIT_WORK_TREE) $_gitworktree
 		cd $pwd
 
 		set status_operation [$::main_status \
-- 
2.54.0.215.g4fe990ec16


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 01/11] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE
  2026-05-22  8:06                       ` Johannes Sixt
@ 2026-05-22 11:54                         ` Mark Levedahl
  2026-05-23  8:18                           ` Johannes Sixt
  0 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-22 11:54 UTC (permalink / raw)
  To: Johannes Sixt, git; +Cc: egg_mushroomcow, bootaina702



On 5/22/26 4:06 AM, Johannes Sixt wrote:
> Am 20.05.26 um 22:24 schrieb Mark Levedahl:
>> 2.54.0.215.g4fe990ec16
>>
>> ------ 8< ------
>> Subject: [PATCH 2/2] git-gui: operate git commands without GIT_WORK_TREE
>>
>> The manual page of the git command states about the --git-dir option:
>>
>>    Specifying the location of the ".git" directory using this option
>>    (or GIT_DIR environment variable) turns off the repository
>>    discovery [...], and tells Git that you are at the top level of
>>    the working tree.
>>
>> Use this to our advantage:
>>
>> - Set GIT_DIR in the environment to the value that was discovered, so
>>   that the invoked git commands operate on the same repository
>>   database that Git GUI uses even after it changes the working
>>   directory.
>>
>> - After changing the working directory to the top level of the working
>>   tree, ensure that GIT_WORK_TREE is not set, because, as per
>>   documentation, all git invocations from then on will assume that the
>>   current working directory is also the top level working tree.
>>
>> - Remove the now obsolete GIT_WORK_TREE dance when subordinate Gitk or
>>   Git GUI are invoked for a submodule.
>
> 2.54.0.215.g4fe990ec16
>
> ------ 8< ------
> Subject: [PATCH 2/2] git-gui: operate git commands without GIT_WORK_TREE
>
> The manual page of the git command states about the --git-dir option:
>
>    Specifying the location of the ".git" directory using this option
>    (or GIT_DIR environment variable) turns off the repository
>    discovery [...], and tells Git that you are at the top level of
>    the working tree.
>
> Use this to our advantage:
>
> - Set GIT_DIR in the environment to the value that was discovered, so
>   that the invoked git commands operate on the same repository
>   database that Git GUI uses even after it changes the working
>   directory.
>
> - After changing the working directory to the top level of the working
>   tree, ensure that GIT_WORK_TREE is not set, because, as per
>   documentation, all git invocations from then on will assume that the
>   current working directory is also the top level working tree.
>
> - Remove the now obsolete GIT_WORK_TREE dance when subordinate Gitk or
>   Git GUI are invoked for a submodule.
> 2.54.0.215.g4fe990ec16
>
> ------ 8< ------
> Subject: [PATCH 2/2] git-gui: operate git commands without GIT_WORK_TREE
>
> The manual page of the git command states about the --git-dir option:
>
>    Specifying the location of the ".git" directory using this option
>    (or GIT_DIR environment variable) turns off the repository
>    discovery [...], and tells Git that you are at the top level of
>    the working tree.
>
> Use this to our advantage:
>
> - Set GIT_DIR in the environment to the value that was discovered, so
>   that the invoked git commands operate on the same repository
>   database that Git GUI uses even after it changes the working
>   directory.
>
> - After changing the working directory to the top level of the working
>   tree, ensure that GIT_WORK_TREE is not set, because, as per
>   documentation, all git invocations from then on will assume that the
>   current working directory is also the top level working tree.
>
> - Remove the now obsolete GIT_WORK_TREE dance when subordinate Gitk or
>   Git GUI are invoked for a submodule.

The manual page is incomplete: if the repository has set core.worktree=/somehere, that is
the root of the worktree and the current directory is always ignored. git rev-parse will
report /somewhere as the answer to --show-toplevel regardless of current directory, even
if inside the gitdir, and even if GIT_DIR is used.

The user can override with GIT_WORK_TREE, and if so we must keep GIT_WORK_TREE in the
environment if it was set. So, an alternative for deciding when to export is

if core.worktree and GIT_WORK_TREE are both set AND GIT_WORK_TREE != core.worktree
    export GIT_WORK_TREE

If the user set GIT_WORK_TREE empty, git rev-parse will throw an error so we would never
to far enough to try to export that.

Mark



^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 01/11] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE
  2026-05-22 11:54                         ` Mark Levedahl
@ 2026-05-23  8:18                           ` Johannes Sixt
  2026-05-23 11:46                             ` Aina Boot
  2026-05-23 16:08                             ` Mark Levedahl
  0 siblings, 2 replies; 134+ messages in thread
From: Johannes Sixt @ 2026-05-23  8:18 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 22.05.26 um 13:54 schrieb Mark Levedahl:
> The manual page is incomplete: if the repository has set core.worktree=/somehere, that is
> the root of the worktree and the current directory is always ignored. git rev-parse will
> report /somewhere as the answer to --show-toplevel regardless of current directory, even
> if inside the gitdir, and even if GIT_DIR is used.
> 
> The user can override with GIT_WORK_TREE, and if so we must keep GIT_WORK_TREE in the
> environment if it was set. [...]

Oh, well, these intricacies! Let's scrap my patch and keep yours.

The other patch that removes cd $_gitworktree from do_gitk should still
be good, I think.

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 01/11] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE
  2026-05-23  8:18                           ` Johannes Sixt
@ 2026-05-23 11:46                             ` Aina Boot
  2026-05-23 16:08                             ` Mark Levedahl
  1 sibling, 0 replies; 134+ messages in thread
From: Aina Boot @ 2026-05-23 11:46 UTC (permalink / raw)
  To: Johannes Sixt, Mark Levedahl; +Cc: Shroom Moo, git

On 5/23/26 8:19 AM, Johannes Sixt wrote:
> The other patch that removes cd $_gitworktree from do_gitk should still
> be good, I think.

Agree. It either succeededly set the directory or work without a 
worktree, cd is practically unnecessary. 

Aina

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 01/11] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE
  2026-05-23  8:18                           ` Johannes Sixt
  2026-05-23 11:46                             ` Aina Boot
@ 2026-05-23 16:08                             ` Mark Levedahl
  1 sibling, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-23 16:08 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: egg_mushroomcow, bootaina702, git



On 5/23/26 4:18 AM, Johannes Sixt wrote:
> Am 22.05.26 um 13:54 schrieb Mark Levedahl:
>> The manual page is incomplete: if the repository has set core.worktree=/somehere, that is
>> the root of the worktree and the current directory is always ignored. git rev-parse will
>> report /somewhere as the answer to --show-toplevel regardless of current directory, even
>> if inside the gitdir, and even if GIT_DIR is used.
>>
>> The user can override with GIT_WORK_TREE, and if so we must keep GIT_WORK_TREE in the
>> environment if it was set. [...]
> Oh, well, these intricacies! Let's scrap my patch and keep yours.
>
> The other patch that removes cd $_gitworktree from do_gitk should still
> be good, I think.
>
> -- Hannes
>
Removing cd $_gitworktree should be ok, we are already in that directory, or don't have a
worktree, and I did that once myself before dropping it as I don't really understand
do_gitk. It should not change any behavior. So, go ahead and add it wherever you wish.

But, I don't understand unsetting GIT_DIR and GIT_WORK_TREE for gitk. If we needed them
for git in the super module, we need them for submodules as well, but have no idea how to
adjust them. Simply unsetting them cannot be right. Out of scope for me. But, there are
some dragons lurking around this proc.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v2 02/11] git-gui: return status from choose_repository::pick
  2026-05-20 20:23                   ` [PATCH v2 " Mark Levedahl
  2026-05-20 20:24                     ` [PATCH v2 01/11] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE Mark Levedahl
@ 2026-05-20 20:24                     ` Mark Levedahl
  2026-05-22  8:18                       ` Johannes Sixt
  2026-05-20 20:24                     ` [PATCH v2 03/11] git-gui: use --absolute-git-dir Mark Levedahl
                                       ` (10 subsequent siblings)
  12 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-20 20:24 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

The repository picker (choose_repository::pick) on success always
returns with the current directory at the root of the selected worktree,
and with the global variable _gitdir holding the name of the git
repository, possibly as a relative path. On failure, _gitdir = {}. If
the selection was from the "recent" list, no validation has occurred.

There are too many side effects in this interface. Note that the picker
only supports worktrees with a .git entry in the worktree root, so git
repository and worktree discovery will work starting in the current
directory on return. So, let's change pick to return a 0/1 value, 1
meaning a worktreee + repo was selected and the current directory is the
worktree root, and leave validation and setting of _gitdir,
_gitworktree, and _prefix to the caller. Note: pick actually does not
return if something was not selected, rather it terminates git-gui.
But, let's pretend at the call site that pick returns 0/false instead.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh                | 10 ++++++++--
 lib/choose_repository.tcl | 21 ++++++++-------------
 2 files changed, 16 insertions(+), 15 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 4ba25da7b6..4a736190a9 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1151,10 +1151,16 @@ if {[catch {
 	} err]} {
 	load_config 1
 	apply_config
-	choose_repository::pick
-	if {![file isdirectory $_gitdir]} {
+	if {![choose_repository::pick]} {
 		exit 1
 	}
+	if {[catch {
+		set _gitdir [git rev-parse --git-dir]
+	} err]} {
+		catch {wm withdraw .}
+		error_popup [strcat [mc "Unusable repo/worktree:"] " [pwd] "\n\n$err"]
+	}
+	set _prefix {}
 	set picked 1
 }
 
diff --git a/lib/choose_repository.tcl b/lib/choose_repository.tcl
index 7e1462a20c..4b06afee93 100644
--- a/lib/choose_repository.tcl
+++ b/lib/choose_repository.tcl
@@ -15,7 +15,7 @@ field w_recentlist ; # Listbox containing recent repositories
 field w_localpath  ; # Entry widget bound to local_path
 
 field done              0 ; # Finished picking the repository?
-field clone_ok      false ; # clone succeeeded
+field pick_ok           0 ; # true if repo pick/clone succeeded
 field local_path       {} ; # Where this repository is locally
 field origin_url       {} ; # Where we are cloning from
 field origin_name  origin ; # What we shall call 'origin'
@@ -220,6 +220,8 @@ constructor pick {} {
 	if {$top eq {.}} {
 		eval destroy [winfo children $top]
 	}
+
+	return $pick_ok
 }
 
 method _center {} {
@@ -327,8 +329,7 @@ method _git_init {} {
 	}
 
 	_append_recentrepos [pwd]
-	set ::_gitdir .git
-	set ::_prefix {}
+	set pick_ok 1
 	return 1
 }
 
@@ -409,6 +410,7 @@ method _do_new2 {} {
 	if {![_git_init $this]} {
 		return
 	}
+	set pick_ok 1
 	set done 1
 }
 
@@ -621,7 +623,7 @@ method _do_clone2 {} {
 	}
 
 	tkwait variable @done
-	if {!$clone_ok} {
+	if {!$pick_ok} {
 		error_popup [mc "Clone failed."]
 		return
 	}
@@ -632,18 +634,12 @@ method _do_clone2_done {ok} {
 	if {$ok} {
 		if {[catch {
 			cd $local_path
-			set ::_gitdir .git
-			set ::_prefix {}
 			_append_recentrepos [pwd]
 		} err]} {
 			set ok 0
 		}
 	}
-	if {!$ok} {
-		set ::_gitdir {}
-		set ::_prefix {}
-	}
-	set clone_ok $ok
+	set pick_ok $ok
 	set done 1
 }
 
@@ -721,8 +717,7 @@ method _do_open2 {} {
 	}
 
 	_append_recentrepos [pwd]
-	set ::_gitdir $actualgit
-	set ::_prefix {}
+	set pick_ok 1
 	set done 1
 }
 
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 02/11] git-gui: return status from choose_repository::pick
  2026-05-20 20:24                     ` [PATCH v2 02/11] git-gui: return status from choose_repository::pick Mark Levedahl
@ 2026-05-22  8:18                       ` Johannes Sixt
  0 siblings, 0 replies; 134+ messages in thread
From: Johannes Sixt @ 2026-05-22  8:18 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 20.05.26 um 22:24 schrieb Mark Levedahl:
> The repository picker (choose_repository::pick) on success always
> returns with the current directory at the root of the selected worktree,
> and with the global variable _gitdir holding the name of the git
> repository, possibly as a relative path. On failure, _gitdir = {}. If
> the selection was from the "recent" list, no validation has occurred.
> 
> There are too many side effects in this interface. Note that the picker
> only supports worktrees with a .git entry in the worktree root, so git
> repository and worktree discovery will work starting in the current
> directory on return. So, let's change pick to return a 0/1 value, 1
> meaning a worktreee + repo was selected and the current directory is the
> worktree root, and leave validation and setting of _gitdir,
> _gitworktree, and _prefix to the caller.

While the removal of side-effects from the picker is very much desired,
the new return value sounds over-engineered at this point, in particular
due to this note:

> Note: pick actually does not
> return if something was not selected, rather it terminates git-gui.
> But, let's pretend at the call site that pick returns 0/false instead.

If we need the return value later, let's postpone that part of this
commit until then.

> diff --git a/git-gui.sh b/git-gui.sh
> index 4ba25da7b6..4a736190a9 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -1151,10 +1151,16 @@ if {[catch {
>  	} err]} {
>  	load_config 1
>  	apply_config
> -	choose_repository::pick
> -	if {![file isdirectory $_gitdir]} {
> +	if {![choose_repository::pick]} {
>  		exit 1
>  	}
> +	if {[catch {
> +		set _gitdir [git rev-parse --git-dir]
> +	} err]} {
> +		catch {wm withdraw .}
> +		error_popup [strcat [mc "Unusable repo/worktree:"] " [pwd] "\n\n$err"]

There's something wrong with the quotes here, and an 'exit 1' is missing.

> +	}
> +	set _prefix {}
>  	set picked 1
>  }
-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v2 03/11] git-gui: use --absolute-git-dir
  2026-05-20 20:23                   ` [PATCH v2 " Mark Levedahl
  2026-05-20 20:24                     ` [PATCH v2 01/11] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE Mark Levedahl
  2026-05-20 20:24                     ` [PATCH v2 02/11] git-gui: return status from choose_repository::pick Mark Levedahl
@ 2026-05-20 20:24                     ` Mark Levedahl
  2026-05-22  8:25                       ` Johannes Sixt
  2026-05-20 20:24                     ` [PATCH v2 04/11] git-gui: use rev-parse exclusively to find a repository Mark Levedahl
                                       ` (9 subsequent siblings)
  12 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-20 20:24 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

git-gui uses git rev-parse --git-dir to get the pathname of the
discovered git repository. The returned value can be relative, and is
'.' if the current directory is the top of the repository directory
itself.  git-gui has code to change '.' to [pwd] in this case so that
subsequent logic runs.

But, git rev-parse supports --absolute-git-dir from fac60b8925
("rev-parse: add option for absolute or relative path formatting",
2020-12-13), and included in git 2.31. git-gui requires git >= 2.36, so
this more useful form is always available. Use --absolute-git-dir to
always get an absolute path, avoiding the need for other checks, and
delete the now unneeded code to fix a relative _gitdir.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 11 ++---------
 1 file changed, 2 insertions(+), 9 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 4a736190a9..233c975786 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1146,7 +1146,7 @@ if {[catch {
 	&& [catch {
 		# beware that from the .git dir this sets _gitdir to .
 		# and _prefix to the empty string
-		set _gitdir [git rev-parse --git-dir]
+		set _gitdir [git rev-parse --absolute-git-dir]
 		set _prefix [git rev-parse --show-prefix]
 	} err]} {
 	load_config 1
@@ -1155,7 +1155,7 @@ if {[catch {
 		exit 1
 	}
 	if {[catch {
-		set _gitdir [git rev-parse --git-dir]
+		set _gitdir [git rev-parse --absolute-git-dir]
 	} err]} {
 		catch {wm withdraw .}
 		error_popup [strcat [mc "Unusable repo/worktree:"] " [pwd] "\n\n$err"]
@@ -1175,13 +1175,6 @@ if {$hashalgorithm eq "sha1"} {
 	exit 1
 }
 
-# we expand the _gitdir when it's just a single dot (i.e. when we're being
-# run from the .git dir itself) lest the routines to find the worktree
-# get confused
-if {$_gitdir eq "."} {
-	set _gitdir [pwd]
-}
-
 if {![file isdirectory $_gitdir]} {
 	catch {wm withdraw .}
 	error_popup [strcat [mc "Git directory not found:"] "\n\n$_gitdir"]
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 03/11] git-gui: use --absolute-git-dir
  2026-05-20 20:24                     ` [PATCH v2 03/11] git-gui: use --absolute-git-dir Mark Levedahl
@ 2026-05-22  8:25                       ` Johannes Sixt
  0 siblings, 0 replies; 134+ messages in thread
From: Johannes Sixt @ 2026-05-22  8:25 UTC (permalink / raw)
  To: Mark Levedahl, git; +Cc: egg_mushroomcow, bootaina702

Am 20.05.26 um 22:24 schrieb Mark Levedahl:
> git-gui uses git rev-parse --git-dir to get the pathname of the
> discovered git repository. The returned value can be relative, and is
> '.' if the current directory is the top of the repository directory
> itself.  git-gui has code to change '.' to [pwd] in this case so that
> subsequent logic runs.
> 
> But, git rev-parse supports --absolute-git-dir from fac60b8925
> ("rev-parse: add option for absolute or relative path formatting",
> 2020-12-13), and included in git 2.31. git-gui requires git >= 2.36, so
> this more useful form is always available. Use --absolute-git-dir to
> always get an absolute path, avoiding the need for other checks, and
> delete the now unneeded code to fix a relative _gitdir.

Very good!

> 
> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
> ---
>  git-gui.sh | 11 ++---------
>  1 file changed, 2 insertions(+), 9 deletions(-)
> 
> diff --git a/git-gui.sh b/git-gui.sh
> index 4a736190a9..233c975786 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -1146,7 +1146,7 @@ if {[catch {
>  	&& [catch {
>  		# beware that from the .git dir this sets _gitdir to .
>  		# and _prefix to the empty string

Note that the comment above needs some adjustment as well.

> -		set _gitdir [git rev-parse --git-dir]
> +		set _gitdir [git rev-parse --absolute-git-dir]
>  		set _prefix [git rev-parse --show-prefix]
>  	} err]} {
>  	load_config 1
> @@ -1155,7 +1155,7 @@ if {[catch {
>  		exit 1
>  	}
>  	if {[catch {
> -		set _gitdir [git rev-parse --git-dir]
> +		set _gitdir [git rev-parse --absolute-git-dir]
>  	} err]} {
>  		catch {wm withdraw .}
>  		error_popup [strcat [mc "Unusable repo/worktree:"] " [pwd] "\n\n$err"]
> @@ -1175,13 +1175,6 @@ if {$hashalgorithm eq "sha1"} {
>  	exit 1
>  }
>  
> -# we expand the _gitdir when it's just a single dot (i.e. when we're being
> -# run from the .git dir itself) lest the routines to find the worktree
> -# get confused
> -if {$_gitdir eq "."} {
> -	set _gitdir [pwd]
> -}
> -
>  if {![file isdirectory $_gitdir]} {
>  	catch {wm withdraw .}
>  	error_popup [strcat [mc "Git directory not found:"] "\n\n$_gitdir"]

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v2 04/11] git-gui: use rev-parse exclusively to find a repository
  2026-05-20 20:23                   ` [PATCH v2 " Mark Levedahl
                                       ` (2 preceding siblings ...)
  2026-05-20 20:24                     ` [PATCH v2 03/11] git-gui: use --absolute-git-dir Mark Levedahl
@ 2026-05-20 20:24                     ` Mark Levedahl
  2026-05-22  8:46                       ` Johannes Sixt
  2026-05-20 20:24                     ` [PATCH v2 05/11] git-gui: simplify [is_bare] to report if a worktree is known Mark Levedahl
                                       ` (8 subsequent siblings)
  12 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-20 20:24 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

git-gui attempts to use env(GIT_DIR) directly as the git repository,
accepting GIT_DIR if it is a directory. Only if that fails is git
rev-parse used to discover the repository.  But, this avoids all of
git-core's validity checking on a repository, thus possibly deferring an
error to a later step, possibly unexpected. Repository validation should
be part of initial setup so that later processing does not need error
trapping for configuration errors.

Let's just invoke rev-parse so all error checking is done.

While here, let's cleanup the error handling.

Stop if an error occurs and the user set GIT_DIR or GIT_WORK_TREE.
Use of either or both of those variables is supported by git, but their
use also means the user has taken responsibility that they are correct,
so a failure is something the user must address.

Otherwise on error, continue the existing behavior and show the
repository picker. But, let's move the possible invocation of
repository_chooser::pick to a separate code block. This permits adding
separate conditions on using pick indepent of repository discovery, and
will be exploited later in the series.  Note that the picker always
returns with the current directory in the root of a worktree with the
git repository is in the .git subdirectory.  The variable "picked" is
used by git-gui to automatically execute the "Explore Working Copy" menu
item after the repository picker is run.  This is controlled by config
variable gui.autoexplore, and happens after all discovery is complete.

Remove a later check on whether _gitdir is a directory: that code
cannot be reached without rev-parse already validating the repository.

_prefix should not be set before worktree discovery: the prefix is only
known after the worktree is found, and at this point we have only
discovered the repository. This is true even when running the repository
picker: that option provides a list of prior selections, and does no
validation on the list beyond checking that the directories exist.  For
now, just initialize _prefix along with other global variables.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 48 +++++++++++++++++++++++++++++++++---------------
 1 file changed, 33 insertions(+), 15 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 233c975786..c61a6cbd8f 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -374,6 +374,7 @@ set _gitdir {}
 set _gitworktree {}
 set _isbare {}
 set _githtmldir {}
+set _prefix {}
 set _reponame {}
 set _shellpath {@@SHELL_PATH@@}
 
@@ -1122,6 +1123,24 @@ unset argv0dir
 ##
 ## repository setup
 
+proc is_gitvars_error {err} {
+	set havevars 0
+	set GIT_DIR {}
+	set GIT_WORK_TREE {}
+	catch {set GIT_DIR $::env(GIT_DIR); set havevars 1}
+	catch {set GIT_WORK_TREE $::env(GIT_WORK_TREE) ; set havevars 1}
+
+	if {$havevars} {
+		catch {wm withdraw .}
+		error_popup [strcat [mc "Invalid configuration:"] \
+		   "\n" "GIT_DIR: " $GIT_DIR \
+		   "\n" "GIT_WORK_TREE: " $GIT_WORK_TREE \
+			"\n\n$err"]
+		return 1
+	}
+	return 0
+}
+
 proc set_gitdir_vars {} {
 	global _gitdir _gitworktree env
 	if {$_gitdir ne {}} {
@@ -1138,17 +1157,22 @@ proc unset_gitdir_vars {} {
 	catch {unset env(GIT_WORK_TREE)}
 }
 
-set picked 0
-if {[catch {
-		set _gitdir $env(GIT_DIR)
-		set _prefix {}
-		}]
-	&& [catch {
-		# beware that from the .git dir this sets _gitdir to .
-		# and _prefix to the empty string
+# find repository.
+set _gitdir {}
+if {$_gitdir eq {}} {
+	if {[catch {
 		set _gitdir [git rev-parse --absolute-git-dir]
-		set _prefix [git rev-parse --show-prefix]
 	} err]} {
+		if {[is_gitvars_error $err]} {
+			exit 1
+		}
+		set _gitdir {}
+	}
+}
+
+set picked 0
+if {$_gitdir eq {}} {
+	unset_gitdir_vars
 	load_config 1
 	apply_config
 	if {![choose_repository::pick]} {
@@ -1160,7 +1184,6 @@ if {[catch {
 		catch {wm withdraw .}
 		error_popup [strcat [mc "Unusable repo/worktree:"] " [pwd] "\n\n$err"]
 	}
-	set _prefix {}
 	set picked 1
 }
 
@@ -1175,11 +1198,6 @@ if {$hashalgorithm eq "sha1"} {
 	exit 1
 }
 
-if {![file isdirectory $_gitdir]} {
-	catch {wm withdraw .}
-	error_popup [strcat [mc "Git directory not found:"] "\n\n$_gitdir"]
-	exit 1
-}
 # _gitdir exists, so try loading the config
 load_config 0
 apply_config
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 04/11] git-gui: use rev-parse exclusively to find a repository
  2026-05-20 20:24                     ` [PATCH v2 04/11] git-gui: use rev-parse exclusively to find a repository Mark Levedahl
@ 2026-05-22  8:46                       ` Johannes Sixt
  2026-05-22 12:04                         ` Mark Levedahl
  2026-05-22 23:00                         ` Mark Levedahl
  0 siblings, 2 replies; 134+ messages in thread
From: Johannes Sixt @ 2026-05-22  8:46 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 20.05.26 um 22:24 schrieb Mark Levedahl:
> git-gui attempts to use env(GIT_DIR) directly as the git repository,
> accepting GIT_DIR if it is a directory. Only if that fails is git
> rev-parse used to discover the repository.  But, this avoids all of
> git-core's validity checking on a repository, thus possibly deferring an
> error to a later step, possibly unexpected. Repository validation should
> be part of initial setup so that later processing does not need error
> trapping for configuration errors.
> 
> Let's just invoke rev-parse so all error checking is done.
> 
> While here, let's cleanup the error handling.
> 
> Stop if an error occurs and the user set GIT_DIR or GIT_WORK_TREE.
> Use of either or both of those variables is supported by git, but their
> use also means the user has taken responsibility that they are correct,
> so a failure is something the user must address.

Very much so!

> 
> Otherwise on error, continue the existing behavior and show the
> repository picker. But, let's move the possible invocation of
> repository_chooser::pick to a separate code block. This permits adding
> separate conditions on using pick indepent of repository discovery, and
> will be exploited later in the series.  Note that the picker always
> returns with the current directory in the root of a worktree with the
> git repository is in the .git subdirectory.  The variable "picked" is
> used by git-gui to automatically execute the "Explore Working Copy" menu
> item after the repository picker is run.  This is controlled by config
> variable gui.autoexplore, and happens after all discovery is complete.
> 
> Remove a later check on whether _gitdir is a directory: that code
> cannot be reached without rev-parse already validating the repository.
> 
> _prefix should not be set before worktree discovery: the prefix is only
> known after the worktree is found, and at this point we have only
> discovered the repository.

Sorry, but I cannot agree with "prefix is only known after the worktree
is found". The prefix is a property that can be known even if we haven't
asked where the top-level of the working tree is. See more below.

> This is true even when running the repository
> picker: that option provides a list of prior selections, and does no
> validation on the list beyond checking that the directories exist.  For
> now, just initialize _prefix along with other global variables.
> 
> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
> ---
>  git-gui.sh | 48 +++++++++++++++++++++++++++++++++---------------
>  1 file changed, 33 insertions(+), 15 deletions(-)
> 
> diff --git a/git-gui.sh b/git-gui.sh
> index 233c975786..c61a6cbd8f 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -374,6 +374,7 @@ set _gitdir {}
>  set _gitworktree {}
>  set _isbare {}
>  set _githtmldir {}
> +set _prefix {}
>  set _reponame {}
>  set _shellpath {@@SHELL_PATH@@}
>  
> @@ -1122,6 +1123,24 @@ unset argv0dir
>  ##
>  ## repository setup
>  
> +proc is_gitvars_error {err} {
> +	set havevars 0
> +	set GIT_DIR {}
> +	set GIT_WORK_TREE {}
> +	catch {set GIT_DIR $::env(GIT_DIR); set havevars 1}
> +	catch {set GIT_WORK_TREE $::env(GIT_WORK_TREE) ; set havevars 1}
> +
> +	if {$havevars} {
> +		catch {wm withdraw .}
> +		error_popup [strcat [mc "Invalid configuration:"] \
> +		   "\n" "GIT_DIR: " $GIT_DIR \
> +		   "\n" "GIT_WORK_TREE: " $GIT_WORK_TREE \
> +			"\n\n$err"]
> +		return 1
> +	}
> +	return 0
> +}
> +
>  proc set_gitdir_vars {} {
>  	global _gitdir _gitworktree env
>  	if {$_gitdir ne {}} {
> @@ -1138,17 +1157,22 @@ proc unset_gitdir_vars {} {
>  	catch {unset env(GIT_WORK_TREE)}
>  }
>  
> -set picked 0
> -if {[catch {
> -		set _gitdir $env(GIT_DIR)
> -		set _prefix {}
> -		}]
> -	&& [catch {
> -		# beware that from the .git dir this sets _gitdir to .
> -		# and _prefix to the empty string
> +# find repository.
> +set _gitdir {}
> +if {$_gitdir eq {}} {
> +	if {[catch {
>  		set _gitdir [git rev-parse --absolute-git-dir]
> -		set _prefix [git rev-parse --show-prefix]

You cannot leave the _prefix empty, because it breaks `git gui browser
master dir` when invoked from a subdirectory of the working tree.

This line must remain. I see that you add it back in later patch. There
may be some motivation to move prefix discovery, but there is no
motivation to remove it at this point.

>  	} err]} {
> +		if {[is_gitvars_error $err]} {
> +			exit 1
> +		}
> +		set _gitdir {}

BTW, this line would only be needed if the 'set _prefix' line above stays.

> +	}
> +}
> +
> +set picked 0
> +if {$_gitdir eq {}} {
> +	unset_gitdir_vars
>  	load_config 1
>  	apply_config
>  	if {![choose_repository::pick]} {
> @@ -1160,7 +1184,6 @@ if {[catch {
>  		catch {wm withdraw .}
>  		error_popup [strcat [mc "Unusable repo/worktree:"] " [pwd] "\n\n$err"]
>  	}
> -	set _prefix {}
>  	set picked 1
>  }
>  
> @@ -1175,11 +1198,6 @@ if {$hashalgorithm eq "sha1"} {
>  	exit 1
>  }
>  
> -if {![file isdirectory $_gitdir]} {
> -	catch {wm withdraw .}
> -	error_popup [strcat [mc "Git directory not found:"] "\n\n$_gitdir"]
> -	exit 1
> -}
>  # _gitdir exists, so try loading the config
>  load_config 0
>  apply_config

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 04/11] git-gui: use rev-parse exclusively to find a repository
  2026-05-22  8:46                       ` Johannes Sixt
@ 2026-05-22 12:04                         ` Mark Levedahl
  2026-05-22 23:00                         ` Mark Levedahl
  1 sibling, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-22 12:04 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: egg_mushroomcow, bootaina702, git



On 5/22/26 4:46 AM, Johannes Sixt wrote:
> Sorry, but I cannot agree with "prefix is only known after the worktree
> is found". The prefix is a property that can be known even if we haven't
> asked where the top-level of the working tree is. See more below.
Rewording is necessary, but:
This patch already adds
set _prefix {}
to the globals init block.


_prefix can only be non-empty if normal gitdir / gitworktree discovery (without GIT_DIR)
runs and finds the current directory is a descendent of the worktree root. That step comes
later, even if the picker is run.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 04/11] git-gui: use rev-parse exclusively to find a repository
  2026-05-22  8:46                       ` Johannes Sixt
  2026-05-22 12:04                         ` Mark Levedahl
@ 2026-05-22 23:00                         ` Mark Levedahl
  1 sibling, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-22 23:00 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: egg_mushroomcow, bootaina702, git

On 5/22/26 4:46 AM, Johannes Sixt wrote:
> Am 20.05.26 um 22:24 schrieb Mark Levedahl:

> 
> Sorry, but I cannot agree with "prefix is only known after the worktree
> is found". The prefix is a property that can be known even if we haven't
> asked where the top-level of the working tree is. See more below.
> 
>> This is true even when running the repository
>> picker: that option provides a list of prior selections, and does no
>> validation on the list beyond checking that the directories exist.  For
>> now, just initialize _prefix along with other global variables.
>>

> 
> You cannot leave the _prefix empty, because it breaks `git gui browser
> master dir` when invoked from a subdirectory of the working tree.
> 
> This line must remain. I see that you add it back in later patch. There
> may be some motivation to move prefix discovery, but there is no
> motivation to remove it at this point.



Never mind, I confused myself on the process. THis patch has only affected repo discovery,
worktree discovery is later so should be left untouched here.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v2 05/11] git-gui: simplify [is_bare] to report if a worktree is known
  2026-05-20 20:23                   ` [PATCH v2 " Mark Levedahl
                                       ` (3 preceding siblings ...)
  2026-05-20 20:24                     ` [PATCH v2 04/11] git-gui: use rev-parse exclusively to find a repository Mark Levedahl
@ 2026-05-20 20:24                     ` Mark Levedahl
  2026-05-20 20:24                     ` [PATCH v2 06/11] git-gui: use git rev-parse for worktree discovery Mark Levedahl
                                       ` (7 subsequent siblings)
  12 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-20 20:24 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

git-gui includes proc is_bare, used in several places to make decisions
on whether a worktree exists, but also in discovery to tell if a
worktree can be supported.

But, is_bare is out of date with regard to multiple worktrees, safe
repository guards, and possibly other relevant features known to git
rev-parse. Also, is_bare caches its result on the first call, so is not
useful if a later step in the discovery process finds a worktree.

So, simplify is_bare to report whether git-gui has a worktree or is
working only from a repository.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 25 +------------------------
 1 file changed, 1 insertion(+), 24 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index c61a6cbd8f..936c309e59 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -372,7 +372,6 @@ if {[tk windowingsystem] eq "aqua"} {
 set _appname {Git Gui}
 set _gitdir {}
 set _gitworktree {}
-set _isbare {}
 set _githtmldir {}
 set _prefix {}
 set _reponame {}
@@ -524,29 +523,7 @@ proc get_config {name} {
 }
 
 proc is_bare {} {
-	global _isbare
-	global _gitdir
-	global _gitworktree
-
-	if {$_isbare eq {}} {
-		if {[catch {
-			set _bare [git rev-parse --is-bare-repository]
-			switch  -- $_bare {
-			true { set _isbare 1 }
-			false { set _isbare 0}
-			default { throw }
-			}
-		}]} {
-			if {[is_config_true core.bare]
-				|| ($_gitworktree eq {}
-					&& [lindex [file split $_gitdir] end] ne {.git})} {
-				set _isbare 1
-			} else {
-				set _isbare 0
-			}
-		}
-	}
-	return $_isbare
+	return [expr {$::_gitworktree eq {}}]
 }
 
 ######################################################################
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* [PATCH v2 06/11] git-gui: use git rev-parse for worktree discovery
  2026-05-20 20:23                   ` [PATCH v2 " Mark Levedahl
                                       ` (4 preceding siblings ...)
  2026-05-20 20:24                     ` [PATCH v2 05/11] git-gui: simplify [is_bare] to report if a worktree is known Mark Levedahl
@ 2026-05-20 20:24                     ` Mark Levedahl
  2026-05-23 13:26                       ` Johannes Sixt
  2026-05-20 20:24                     ` [PATCH v2 07/11] git-gui: try harder to find worktree from gitdir Mark Levedahl
                                       ` (6 subsequent siblings)
  12 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-20 20:24 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

git gui uses a combination of tcl code and git invocations to determine
the worktree and the location with respect to the worktree root
(_prefix). But, git rev-parse provides all of this information directly,
and assures full error and configuration checking are done by git
itself. The entirety of discovery in normal configurations involves

	git rev-parse --show-toplevel (gets worktree root)
	git rev-parse --show-prefix (shows location wrt the root)

An error thrown on either of these lines means the worktree discovered
by git is unusable, or git did not discover a worktree because the
current directory is inside the repository. If the user has defined
GIT_DIR or GIT_WORK_TREE, this is a user configuration error and git-gui
should stop.

Otherwise, the blame or browser subcommands can be used without a
worktree.

A separate error might occur when changing to the root of the discovered
worktree. The cause would be file system related and completely outside
of git's control. So, the final "cd $worktree_root" is separately
trapped.

Discovery of the repository and the worktree must be guarded to trap
errors: the intent is that any configuration problems are caught during
discovery, and later processing need not include error trapping and
recovery. So, move all worktree discovery code to be immediately after
repository discovery.

This does move configuration loading to occur after worktree discovery
rather than before. None of the code executed in worktree discovery has
any option controlled by a git-gui configuration variable, so no impact
is expected. git itself will always read the repository configuration,
including worktree specific configuration data if that exists, so this
is unaffected by when git-gui loads its own config data, and we cannot
be sure the full worktree dependent configuration can be loaded before
full discovery is complete.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 64 +++++++++++++++++++++++++-----------------------------
 1 file changed, 30 insertions(+), 34 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 936c309e59..8fe25fe188 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1164,6 +1164,36 @@ if {$_gitdir eq {}} {
 	set picked 1
 }
 
+# find worktree, continue without if not required
+if {[catch {
+	set _gitworktree [git rev-parse --show-toplevel]
+	set _prefix [git rev-parse --show-prefix]
+} err]} {
+	if {[is_gitvars_error $err]} {
+		exit 1
+	}
+	set _gitworktree {}
+	set _prefix {}
+}
+
+if {![is_bare]} {
+	if {[catch {
+		cd $_gitworktree
+	} err]} {
+		catch {wm withdraw .}
+		error_popup [strcat [mc "Cannot change to discovered worktree: "] \
+			"$_gitworktree" "\n\n$err"]
+		exit 1;
+	}
+} elseif {![is_enabled bare]} {
+	catch {wm withdraw .}
+	error_popup [strcat [mc "Cannot use bare repository:"] "\n\n" $_gitdir]
+	exit 1
+}
+
+# repository and worktree config are complete, export them
+set_gitdir_vars
+
 # Use object format as hash algorithm (either "sha1" or "sha256")
 set hashalgorithm [git rev-parse --show-object-format]
 if {$hashalgorithm eq "sha1"} {
@@ -1179,37 +1209,6 @@ if {$hashalgorithm eq "sha1"} {
 load_config 0
 apply_config
 
-set _gitworktree [git rev-parse --show-toplevel]
-
-if {$_prefix ne {}} {
-	if {$_gitworktree eq {}} {
-		regsub -all {[^/]+/} $_prefix ../ cdup
-	} else {
-		set cdup $_gitworktree
-	}
-	if {[catch {cd $cdup} err]} {
-		catch {wm withdraw .}
-		error_popup [strcat [mc "Cannot move to top of working directory:"] "\n\n$err"]
-		exit 1
-	}
-	set _gitworktree [pwd]
-	unset cdup
-} elseif {![is_enabled bare]} {
-	if {[is_bare]} {
-		catch {wm withdraw .}
-		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n$_gitdir"]
-		exit 1
-	}
-	if {$_gitworktree eq {}} {
-		set _gitworktree [file dirname $_gitdir]
-	}
-	if {[catch {cd $_gitworktree} err]} {
-		catch {wm withdraw .}
-		error_popup [strcat [mc "No working directory"] " $_gitworktree:\n\n$err"]
-		exit 1
-	}
-	set _gitworktree [pwd]
-}
 set _reponame [file split [file normalize $_gitdir]]
 if {[lindex $_reponame end] eq {.git}} {
 	set _reponame [lindex $_reponame end-1]
@@ -1217,9 +1216,6 @@ if {[lindex $_reponame end] eq {.git}} {
 	set _reponame [lindex $_reponame end]
 }
 
-# Export the final paths
-set_gitdir_vars
-
 ######################################################################
 ##
 ## global init
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 06/11] git-gui: use git rev-parse for worktree discovery
  2026-05-20 20:24                     ` [PATCH v2 06/11] git-gui: use git rev-parse for worktree discovery Mark Levedahl
@ 2026-05-23 13:26                       ` Johannes Sixt
  0 siblings, 0 replies; 134+ messages in thread
From: Johannes Sixt @ 2026-05-23 13:26 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 20.05.26 um 22:24 schrieb Mark Levedahl:
> git gui uses a combination of tcl code and git invocations to determine
> the worktree and the location with respect to the worktree root
> (_prefix). But, git rev-parse provides all of this information directly,
> and assures full error and configuration checking are done by git
> itself. The entirety of discovery in normal configurations involves
> 
> 	git rev-parse --show-toplevel (gets worktree root)
> 	git rev-parse --show-prefix (shows location wrt the root)
> 
> An error thrown on either of these lines means the worktree discovered
> by git is unusable, or git did not discover a worktree because the
> current directory is inside the repository. If the user has defined
> GIT_DIR or GIT_WORK_TREE, this is a user configuration error and git-gui
> should stop.
> 
> Otherwise, the blame or browser subcommands can be used without a
> worktree.
> 
> A separate error might occur when changing to the root of the discovered
> worktree. The cause would be file system related and completely outside
> of git's control. So, the final "cd $worktree_root" is separately
> trapped.
> 
> Discovery of the repository and the worktree must be guarded to trap
> errors: the intent is that any configuration problems are caught during
> discovery, and later processing need not include error trapping and
> recovery. So, move all worktree discovery code to be immediately after
> repository discovery.
> 
> This does move configuration loading to occur after worktree discovery
> rather than before. None of the code executed in worktree discovery has
> any option controlled by a git-gui configuration variable, so no impact
> is expected. git itself will always read the repository configuration,
> including worktree specific configuration data if that exists, so this
> is unaffected by when git-gui loads its own config data, and we cannot
> be sure the full worktree dependent configuration can be loaded before
> full discovery is complete.

Very good!

When you move code around, please do not apply style changes so that

git show --color-moved --color-moved-ws=allow-indentation-change

can prove that no change was intended.

> 
> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
> ---
>  git-gui.sh | 64 +++++++++++++++++++++++++-----------------------------
>  1 file changed, 30 insertions(+), 34 deletions(-)
> 
> diff --git a/git-gui.sh b/git-gui.sh
> index 936c309e59..8fe25fe188 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -1164,6 +1164,36 @@ if {$_gitdir eq {}} {
>  	set picked 1
>  }
>  
> +# find worktree, continue without if not required
> +if {[catch {
> +	set _gitworktree [git rev-parse --show-toplevel]
> +	set _prefix [git rev-parse --show-prefix]
> +} err]} {
> +	if {[is_gitvars_error $err]} {
> +		exit 1
> +	}
> +	set _gitworktree {}
> +	set _prefix {}
> +}
> +
> +if {![is_bare]} {
> +	if {[catch {
> +		cd $_gitworktree
> +	} err]} {
> +		catch {wm withdraw .}
> +		error_popup [strcat [mc "Cannot change to discovered worktree: "] \
> +			"$_gitworktree" "\n\n$err"]
> +		exit 1;
> +	}
> +} elseif {![is_enabled bare]} {
> +	catch {wm withdraw .}
> +	error_popup [strcat [mc "Cannot use bare repository:"] "\n\n" $_gitdir]
> +	exit 1
> +}
> +
> +# repository and worktree config are complete, export them
> +set_gitdir_vars
> +
>  # Use object format as hash algorithm (either "sha1" or "sha256")
>  set hashalgorithm [git rev-parse --show-object-format]
>  if {$hashalgorithm eq "sha1"} {
> @@ -1179,37 +1209,6 @@ if {$hashalgorithm eq "sha1"} {
>  load_config 0
>  apply_config
>  
> -set _gitworktree [git rev-parse --show-toplevel]
> -
> -if {$_prefix ne {}} {
> -	if {$_gitworktree eq {}} {
> -		regsub -all {[^/]+/} $_prefix ../ cdup
> -	} else {
> -		set cdup $_gitworktree
> -	}
> -	if {[catch {cd $cdup} err]} {
> -		catch {wm withdraw .}
> -		error_popup [strcat [mc "Cannot move to top of working directory:"] "\n\n$err"]
> -		exit 1
> -	}
> -	set _gitworktree [pwd]
> -	unset cdup
> -} elseif {![is_enabled bare]} {
> -	if {[is_bare]} {
> -		catch {wm withdraw .}
> -		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n$_gitdir"]
> -		exit 1
> -	}
> -	if {$_gitworktree eq {}} {
> -		set _gitworktree [file dirname $_gitdir]
> -	}
> -	if {[catch {cd $_gitworktree} err]} {
> -		catch {wm withdraw .}
> -		error_popup [strcat [mc "No working directory"] " $_gitworktree:\n\n$err"]
> -		exit 1
> -	}
> -	set _gitworktree [pwd]
> -}
>  set _reponame [file split [file normalize $_gitdir]]
>  if {[lindex $_reponame end] eq {.git}} {
>  	set _reponame [lindex $_reponame end-1]
> @@ -1217,9 +1216,6 @@ if {[lindex $_reponame end] eq {.git}} {
>  	set _reponame [lindex $_reponame end]
>  }
>  
> -# Export the final paths
> -set_gitdir_vars
> -
>  ######################################################################
>  ##
>  ## global init

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v2 07/11] git-gui: try harder to find worktree from gitdir
  2026-05-20 20:23                   ` [PATCH v2 " Mark Levedahl
                                       ` (5 preceding siblings ...)
  2026-05-20 20:24                     ` [PATCH v2 06/11] git-gui: use git rev-parse for worktree discovery Mark Levedahl
@ 2026-05-20 20:24                     ` Mark Levedahl
  2026-05-21  4:55                       ` Shroom Moo
  2026-05-23 14:06                       ` Johannes Sixt
  2026-05-20 20:24                     ` [PATCH v2 08/11] git-gui: use HEAD as current branch when detached (bug fix) Mark Levedahl
                                       ` (5 subsequent siblings)
  12 siblings, 2 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-20 20:24 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

git-gui, since 87cd09f43e ("git-gui: work from the .git dir",
2010-01-23), has had the intent to allow starting from inside a
repository, then switching to the parent directory if that is a valid
worktree.

This certainly hasn't worked since 2d92ab32fd ("rev-parse: make
--show-toplevel without a worktree an error", 2019-11-19) in git, but
breaking this git-gui feature was unintentional.

There are (at least) 3 cases where the gitdir can tell us where the
worktree is, and we would like all to work:

- core.worktree is set, and points to a valid worktree. This is already
  handled  by git rev-parse --show-toplevel, even when not in the worktree.
  There is nothing more to do in this case.

- the gitdir is embedded in a worktree as subdirectory .git. The parent
  is (or at least should be) a valid worktree. This worked long ago.

- the gitdir is a worktree specific directory (under
  <mainrepo>/worktrees/worktree_name), within which there is a file
  "gitdir" pointing to .git in the worktree. git gui never learned to
  handle this case.

Let's handle the latter two cases. Always check that the discovered
worktree is valid and points to the already discovered gitdir according
to git rev-parse. This avoids issues that may arise because we are
discovering from the gitdir up, rather than the worktree down, and file
system non-posix behavior or misconfiguration of git might cause
confusion.  For instance, a manually moved worktree might not be where
the gitdir points.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 42 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 42 insertions(+)

diff --git a/git-gui.sh b/git-gui.sh
index 8fe25fe188..aeb7ed3548 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1100,6 +1100,41 @@ unset argv0dir
 ##
 ## repository setup
 
+proc find_worktree_from_gitdir {} {
+	# Directory 'parent' of a repository named 'parent/.git' might be the worktree.
+	# Assure parent is a worktree and using the git repository already discovered.
+	# Also, handle case of being in a worktree's gitdir, where file "gitdir" points to
+	# gitlink file .git in the real worktree.
+	set worktree {}
+	if {[file tail $::_gitdir] eq {.git}} {
+		if {[catch {
+			set gitdir_parent [file dirname $::_gitdir]
+			set worktree [git -C $gitdir_parent rev-parse --show-toplevel]
+			set parent_gitdir [git -C $worktree rev-parse --absolute-git-dir]
+			if {$::_gitdir ne $parent_gitdir} {
+				set worktree {}
+			}
+		}]} {
+			set worktree {}
+		}
+	} elseif [file exists {gitdir}] {
+		if {[catch {
+			set fd_gitdir [open {gitdir} {r}]
+			set gitlink_parent [file dirname [read $fd_gitdir]]
+			catch {close $fd_gitdir}
+			set worktree [git -C $gitlink_parent rev-parse --show-toplevel]
+			set parent_gitdir [git -C $worktree rev-parse --absolute-git-dir]
+			if {$::_gitdir ne $parent_gitdir} {
+				set worktree {}
+			}
+		}]} {
+			catch {close $fd_gitdir}
+			set worktree {}
+		}
+	}
+	return $worktree
+}
+
 proc is_gitvars_error {err} {
 	set havevars 0
 	set GIT_DIR {}
@@ -1176,6 +1211,13 @@ if {[catch {
 	set _prefix {}
 }
 
+if {[is_bare]} {
+	# Maybe we are in an embedded or worktree specific gitdir
+	if {[set _gitworktree [find_worktree_from_gitdir]] ne {}} {
+		set _prefix {}
+	}
+}
+
 if {![is_bare]} {
 	if {[catch {
 		cd $_gitworktree
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 07/11] git-gui: try harder to find worktree from gitdir
  2026-05-20 20:24                     ` [PATCH v2 07/11] git-gui: try harder to find worktree from gitdir Mark Levedahl
@ 2026-05-21  4:55                       ` Shroom Moo
  2026-05-21 17:45                         ` Mark Levedahl
  2026-05-23  8:01                         ` Johannes Sixt
  2026-05-23 14:06                       ` Johannes Sixt
  1 sibling, 2 replies; 134+ messages in thread
From: Shroom Moo @ 2026-05-21  4:55 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: git, Johannes Sixt, Aina Boot

On 5/21/26 4:24 AM, Mark Levedahl wrote:
> +	} elseif [file exists {gitdir}] {
> +		if {[catch {
> +			set fd_gitdir [open {gitdir} {r}]
> +			set gitlink_parent [file dirname [read $fd_gitdir]]
> +			catch {close $fd_gitdir}
> +			set worktree [git -C $gitlink_parent rev-parse --show-toplevel]
> +			set parent_gitdir [git -C $worktree rev-parse --absolute-git-dir]
> +			if {$::_gitdir ne $parent_gitdir} {
> +				set worktree {}
> +			}
> +		}]} {
> +			catch {close $fd_gitdir}
> +			set worktree {}
> +		}
> +	}

There is also an unaddressed issue: 
In [file exists {gitdir}] and [open {gitdir} r], {gitdir} is a 
literal string referring to a file named gitdir in the current 
working directory. However, in the context of a linked worktree 
(created via git worktree add), the actual file path is 
$_gitdir/gitdir (e.g., .git/worktrees/<name>/gitdir). While the 
current working directory could be anywhere (even inside the .git 
directory), $_gitdir is an absolute path pointing to that worktree's 
gitdir (e.g., /path/to/main/.git/worktrees/branch). The gitdir file 
resides within the $_gitdir directory and contains a relative path 
like ../../.git/worktrees/branch. The current code logic will never 
locate this file. 

Additionally, [file exists {gitdir}] checks for the gitdir file in 
the current working directory. Since the function has not yet 
switched to $_gitdir when this check runs, it is almost impossible 
to find the file. Consequently, this logic never triggers, preventing 
linked worktrees from being recognized. 

Maybe the identification of linked worktree should not directly look 
for the gitdir file, but should check whether there is a.git file and 
its content points to... /.git/worktrees/... ? Anyways, using the 
literal {gitdir} to search in the current directory lead to risks. 

Shroom

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 07/11] git-gui: try harder to find worktree from gitdir
  2026-05-21  4:55                       ` Shroom Moo
@ 2026-05-21 17:45                         ` Mark Levedahl
  2026-05-22 15:09                           ` Shroom Moo
  2026-05-23  8:01                         ` Johannes Sixt
  1 sibling, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-21 17:45 UTC (permalink / raw)
  To: Shroom Moo; +Cc: git, Johannes Sixt, Aina Boot



On 5/21/26 12:55 AM, Shroom Moo wrote:
> On 5/21/26 4:24 AM, Mark Levedahl wrote:
>> +	} elseif [file exists {gitdir}] {
>> +		if {[catch {
>> +			set fd_gitdir [open {gitdir} {r}]
>> +			set gitlink_parent [file dirname [read $fd_gitdir]]
>> +			catch {close $fd_gitdir}
>> +			set worktree [git -C $gitlink_parent rev-parse --show-toplevel]
>> +			set parent_gitdir [git -C $worktree rev-parse --absolute-git-dir]
>> +			if {$::_gitdir ne $parent_gitdir} {
>> +				set worktree {}
>> +			}
>> +		}]} {
>> +			catch {close $fd_gitdir}
>> +			set worktree {}
>> +		}
>> +	}
> There is also an unaddressed issue: 
> In [file exists {gitdir}] and [open {gitdir} r], {gitdir} is a 
> literal string referring to a file named gitdir in the current 
> working directory. However, in the context of a linked worktree 
> (created via git worktree add), the actual file path is 
> $_gitdir/gitdir (e.g., .git/worktrees/<name>/gitdir). While the 
> current working directory could be anywhere (even inside the .git 
> directory), $_gitdir is an absolute path pointing to that worktree's 
> gitdir (e.g., /path/to/main/.git/worktrees/branch). The gitdir file 
> resides within the $_gitdir directory and contains a relative path 
> like ../../.git/worktrees/branch. The current code logic will never 
> locate this file. 
You have to be in the particular worktree's gitdir for this to work. I there exists
    worktrees/foo
    worktrees/frotz
    worktrees/bar
Which would we expore? The code above must be in foo, frotz, bar

The main worktree is found not from worktrees/*, but from the root of the gitdir.
>
> Additionally, [file exists {gitdir}] checks for the gitdir file in 
> the current working directory. Since the function has not yet 
> switched to $_gitdir when this check runs, it is almost impossible 
> to find the file. Consequently, this logic never triggers, preventing 
> linked worktrees from being recognized. 
>
> Maybe the identification of linked worktree should not directly look 
> for the gitdir file, but should check whether there is a.git file and 
> its content points to... /.git/worktrees/... ? Anyways, using the 
> literal {gitdir} to search in the current directory lead to risks. 
>
> Shroom
>
We cannot get to this code if not inside the gitdir, and if the user set GIT_DIR and/or
GIT_WORK_TREE to do something clever, that either worked or the code already threw an
error. git, without GIT_WORK_TREE set, uses the current directory as the worktree, or the
parent directory containing .git. So, we must be inside the gitdir if this code path gets hit.

Mark

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 07/11] git-gui: try harder to find worktree from gitdir
  2026-05-21 17:45                         ` Mark Levedahl
@ 2026-05-22 15:09                           ` Shroom Moo
  2026-05-22 16:57                             ` Mark Levedahl
  0 siblings, 1 reply; 134+ messages in thread
From: Shroom Moo @ 2026-05-22 15:09 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: git, Johannes Sixt, Aina Boot

On 5/22/26 1:45 AM, Mark Levedahl wrote:
> >
> > Additionally, [file exists {gitdir}] checks for the gitdir file in 
> > the current working directory. Since the function has not yet 
> > switched to $_gitdir when this check runs, it is almost impossible 
> > to find the file. Consequently, this logic never triggers, preventing 
> > linked worktrees from being recognized. 
> >
> > Maybe the identification of linked worktree should not directly look 
> > for the gitdir file, but should check whether there is a.git file and 
> > its content points to... /.git/worktrees/... ? Anyways, using the 
> > literal {gitdir} to search in the current directory lead to risks. 
> >
> > Shroom
> >
> We cannot get to this code if not inside the gitdir, and if the user set GIT_DIR and/or
> GIT_WORK_TREE to do something clever, that either worked or the code already threw an
> error. git, without GIT_WORK_TREE set, uses the current directory as the worktree, or the
> parent directory containing .git. So, we must be inside the gitdir if this code path gets hit.
> 
> Mark

The relative path issue with {gitdir} is indeed difficult to trigger 
in practice. To trigger this problem, the following conditions must 
be met simultaneously: 
- git rev-parse --absolute-git-dir succeeds (the repository is valid). 
- git rev-parse --show-toplevel fails (the working tree is 
undetectable).
- The current directory is not $_gitdir.
- A gitdir file exists under $_gitdir (i.e., the gitdir of a linked 
worktree). 

Unluckily, this occurs in:

```MINGW64 Shell
# Firstly, enter a test folder
mkdir test-main
cd test-main

git init

echo main > file.txt
git add file.txt
git commit -m "initial"

# In feature branch
git branch feature

# Add worktree
git worktree add ../test-feature feature

cd ../
cd test-feature
cat .git

cd ../

mkdir outside
cd outside

export GIT_DIR="/(PREVIOUSPATH)/test-main/.git/worktrees/feature"
unset GIT_WORK_TREE
git gui blame HEAD file.txt --trace
```

```Wish
puts $_gitworktree
```

The current directory is "outside", not the expected linked worktree. 
[file exists {gitdir}] in find_worktree_from_gitdir searched in the 
current directory but failed, resulting in the inability to identify 
linked worktree. The blame mode can still work without worktree. 

Shroom


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 07/11] git-gui: try harder to find worktree from gitdir
  2026-05-22 15:09                           ` Shroom Moo
@ 2026-05-22 16:57                             ` Mark Levedahl
  0 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-22 16:57 UTC (permalink / raw)
  To: Shroom Moo; +Cc: git, Johannes Sixt, Aina Boot



On 5/22/26 11:09 AM, Shroom Moo wrote:
> On 5/22/26 1:45 AM, Mark Levedahl wrote:
>>> Additionally, [file exists {gitdir}] checks for the gitdir file in 
>>> the current working directory. Since the function has not yet 
>>> switched to $_gitdir when this check runs, it is almost impossible 
>>> to find the file. Consequently, this logic never triggers, preventing 
>>> linked worktrees from being recognized. 
>>>
>>> Maybe the identification of linked worktree should not directly look 
>>> for the gitdir file, but should check whether there is a.git file and 
>>> its content points to... /.git/worktrees/... ? Anyways, using the 
>>> literal {gitdir} to search in the current directory lead to risks. 
>>>
>>> Shroom
>>>
>> We cannot get to this code if not inside the gitdir, and if the user set GIT_DIR and/or
>> GIT_WORK_TREE to do something clever, that either worked or the code already threw an
>> error. git, without GIT_WORK_TREE set, uses the current directory as the worktree, or the
>> parent directory containing .git. So, we must be inside the gitdir if this code path gets hit.
>>
>> Mark
> The relative path issue with {gitdir} is indeed difficult to trigger 
> in practice. To trigger this problem, the following conditions must 
> be met simultaneously: 
> - git rev-parse --absolute-git-dir succeeds (the repository is valid). 
> - git rev-parse --show-toplevel fails (the working tree is 
> undetectable).
> - The current directory is not $_gitdir.
> - A gitdir file exists under $_gitdir (i.e., the gitdir of a linked 
> worktree). 
>
> Unluckily, this occurs in:
>
> ```MINGW64 Shell
> # Firstly, enter a test folder
> mkdir test-main
> cd test-main
>
> git init
>
> echo main > file.txt
> git add file.txt
> git commit -m "initial"
>
> # In feature branch
> git branch feature
>
> # Add worktree
> git worktree add ../test-feature feature
>
> cd ../
> cd test-feature
> cat .git
>
> cd ../
>
> mkdir outside
> cd outside
>
> export GIT_DIR="/(PREVIOUSPATH)/test-main/.git/worktrees/feature"
> unset GIT_WORK_TREE
> git gui blame HEAD file.txt --trace
> ```
>
> ```Wish
> puts $_gitworktree
> ```
>
> The current directory is "outside", not the expected linked worktree. 
> [file exists {gitdir}] in find_worktree_from_gitdir searched in the 
> current directory but failed, resulting in the inability to identify 
> linked worktree. The blame mode can still work without worktree. 
>
> Shroom
>
You exported GIT_DIR pointing to a valid gitdir, and as documented git takes your current
directory ("outside") to be the worktree. This is working as expected. Had finding a
worktree failed, git-gui would issue an error message and stop because you defined GIT_DIR.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 07/11] git-gui: try harder to find worktree from gitdir
  2026-05-21  4:55                       ` Shroom Moo
  2026-05-21 17:45                         ` Mark Levedahl
@ 2026-05-23  8:01                         ` Johannes Sixt
  2026-05-23 11:47                           ` Shroom Moo
  1 sibling, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-23  8:01 UTC (permalink / raw)
  To: Shroom Moo; +Cc: git, Aina Boot, Mark Levedahl

Am 21.05.26 um 06:55 schrieb Shroom Moo:
> On 5/21/26 4:24 AM, Mark Levedahl wrote:
>> +	} elseif [file exists {gitdir}] {
>> +		if {[catch {
>> +			set fd_gitdir [open {gitdir} {r}]
>> +			set gitlink_parent [file dirname [read $fd_gitdir]]
>> +			catch {close $fd_gitdir}
>> +			set worktree [git -C $gitlink_parent rev-parse --show-toplevel]
>> +			set parent_gitdir [git -C $worktree rev-parse --absolute-git-dir]
>> +			if {$::_gitdir ne $parent_gitdir} {
>> +				set worktree {}
>> +			}
>> +		}]} {
>> +			catch {close $fd_gitdir}
>> +			set worktree {}
>> +		}
>> +	}

> Additionally, [file exists {gitdir}] checks for the gitdir file in 
> the current working directory. Since the function has not yet 
> switched to $_gitdir when this check runs, it is almost impossible 
> to find the file. Consequently, this logic never triggers, preventing 
> linked worktrees from being recognized. 

I think you are misunderstanding which use-case this code is addressing.
The case can be triggered very easily.

First, the code before the part we see above is intended for the special
case where we start in a .git, where `--show-toplevel` bails out and we
define the worktree to be the directory containing .git.

However, if we start in .git/worktrees/feature, then the code cited
above kicks in, because `--show-toplevel` still bails out,
`--absolute-git-dir` does not end in '.git', but now we have a file
named 'gitdir' in the current directory. In this case, we define (and
this is new with this patch) that the worktree is the one where the
'gitdir' points.

-- Hannes

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 07/11] git-gui: try harder to find worktree from gitdir
  2026-05-23  8:01                         ` Johannes Sixt
@ 2026-05-23 11:47                           ` Shroom Moo
  0 siblings, 0 replies; 134+ messages in thread
From: Shroom Moo @ 2026-05-23 11:47 UTC (permalink / raw)
  To: Mark Levedahl, Johannes Sixt; +Cc: git, Aina Boot

On 5/23/26 4:01 PM, Johannes Sixt wrote:
> Am 21.05.26 um 06:55 schrieb Shroom Moo:
>> On 5/21/26 4:24 AM, Mark Levedahl wrote:
>>> +	} elseif [file exists {gitdir}] {
>>> +		if {[catch {
>>> +			set fd_gitdir [open {gitdir} {r}]
>>> +			set gitlink_parent [file dirname [read $fd_gitdir]]
>>> +			catch {close $fd_gitdir}
>>> +			set worktree [git -C $gitlink_parent rev-parse --show-toplevel]
>>> +			set parent_gitdir [git -C $worktree rev-parse --absolute-git-dir]
>>> +			if {$::_gitdir ne $parent_gitdir} {
>>> +				set worktree {}
>>> +			}
>>> +		}]} {
>>> +			catch {close $fd_gitdir}
>>> +			set worktree {}
>>> +		}
>>> +	}

>> Additionally, [file exists {gitdir}] checks for the gitdir file in 
>> the current working directory. Since the function has not yet 
>> switched to $_gitdir when this check runs, it is almost impossible 
>> to find the file. Consequently, this logic never triggers, preventing 
>> linked worktrees from being recognized. 

> I think you are misunderstanding which use-case this code is addressing.
> The case can be triggered very easily.

> First, the code before the part we see above is intended for the special
> case where we start in a .git, where `--show-toplevel` bails out and we
> define the worktree to be the directory containing .git.
> However, if we start in .git/worktrees/feature, then the code cited
> above kicks in, because `--show-toplevel` still bails out,
> `--absolute-git-dir` does not end in '.git', but now we have a file
> named 'gitdir' in the current directory. In this case, we define (and
> this is new with this patch) that the worktree is the one where the
> 'gitdir' points.
> 
> -- Hannes

I see. The condition is unrelated to this patch. Users should handle 
this case as assigning manually by rule. Indeed we don't need to 
modify it. 

Shroom


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 07/11] git-gui: try harder to find worktree from gitdir
  2026-05-20 20:24                     ` [PATCH v2 07/11] git-gui: try harder to find worktree from gitdir Mark Levedahl
  2026-05-21  4:55                       ` Shroom Moo
@ 2026-05-23 14:06                       ` Johannes Sixt
  2026-05-23 15:33                         ` Mark Levedahl
  1 sibling, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-23 14:06 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 20.05.26 um 22:24 schrieb Mark Levedahl:
> git-gui, since 87cd09f43e ("git-gui: work from the .git dir",
> 2010-01-23), has had the intent to allow starting from inside a
> repository, then switching to the parent directory if that is a valid
> worktree.

I can imagine that this kind of use occurs in "Git GUI here" menu item
of the file explorer on Windows. So, we should resurrect the feature.

> 
> This certainly hasn't worked since 2d92ab32fd ("rev-parse: make
> --show-toplevel without a worktree an error", 2019-11-19) in git, but
> breaking this git-gui feature was unintentional.
> 
> There are (at least) 3 cases where the gitdir can tell us where the
> worktree is, and we would like all to work:
> 
> - core.worktree is set, and points to a valid worktree. This is already
>   handled  by git rev-parse --show-toplevel, even when not in the worktree.
>   There is nothing more to do in this case.
> 
> - the gitdir is embedded in a worktree as subdirectory .git. The parent
>   is (or at least should be) a valid worktree. This worked long ago.
> 
> - the gitdir is a worktree specific directory (under
>   <mainrepo>/worktrees/worktree_name), within which there is a file
>   "gitdir" pointing to .git in the worktree. git gui never learned to
>   handle this case.
> 
> Let's handle the latter two cases. Always check that the discovered
> worktree is valid and points to the already discovered gitdir according
> to git rev-parse. This avoids issues that may arise because we are
> discovering from the gitdir up, rather than the worktree down, and file
> system non-posix behavior or misconfiguration of git might cause
> confusion.  For instance, a manually moved worktree might not be where
> the gitdir points.
> 
> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
> ---
>  git-gui.sh | 42 ++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 42 insertions(+)
> 
> diff --git a/git-gui.sh b/git-gui.sh
> index 8fe25fe188..aeb7ed3548 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -1100,6 +1100,41 @@ unset argv0dir
>  ##
>  ## repository setup
>  
> +proc find_worktree_from_gitdir {} {
> +	# Directory 'parent' of a repository named 'parent/.git' might be the worktree.
> +	# Assure parent is a worktree and using the git repository already discovered.
> +	# Also, handle case of being in a worktree's gitdir, where file "gitdir" points to
> +	# gitlink file .git in the real worktree.
> +	set worktree {}
> +	if {[file tail $::_gitdir] eq {.git}} {
> +		if {[catch {
> +			set gitdir_parent [file dirname $::_gitdir]
> +			set worktree [git -C $gitdir_parent rev-parse --show-toplevel]
> +			set parent_gitdir [git -C $worktree rev-parse --absolute-git-dir]
> +			if {$::_gitdir ne $parent_gitdir} {
> +				set worktree {}

I tried to come up with a situation where we end up here, but couldn't.
When would this happen? If it actually can't happen, I would prefer to
spawn fewer git processes and just take the result of 'file dirname'.

If the code must remain, can we please rename one of gitdir_parent or
parent_gitdir?

> +			}
> +		}]} {
> +			set worktree {}
> +		}
> +	} elseif [file exists {gitdir}] {
> +		if {[catch {
> +			set fd_gitdir [open {gitdir} {r}]
> +			set gitlink_parent [file dirname [read $fd_gitdir]]
> +			catch {close $fd_gitdir}
> +			set worktree [git -C $gitlink_parent rev-parse --show-toplevel]
> +			set parent_gitdir [git -C $worktree rev-parse --absolute-git-dir]

Since worktrees can be messed up quite easily, it looks reasonable to
check whether the worktree points back to the gitdir. (But I haven't
tried to construct a case that passes the check in the next line.)

> +			if {$::_gitdir ne $parent_gitdir} {
> +				set worktree {}
> +			}
> +		}]} {
> +			catch {close $fd_gitdir}
> +			set worktree {}
> +		}
> +	}
> +	return $worktree
> +}
> +
>  proc is_gitvars_error {err} {
>  	set havevars 0
>  	set GIT_DIR {}
> @@ -1176,6 +1211,13 @@ if {[catch {
>  	set _prefix {}
>  }
>  
> +if {[is_bare]} {
> +	# Maybe we are in an embedded or worktree specific gitdir
> +	if {[set _gitworktree [find_worktree_from_gitdir]] ne {}} {
> +		set _prefix {}
> +	}
> +}
> +
>  if {![is_bare]} {
>  	if {[catch {
>  		cd $_gitworktree

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 07/11] git-gui: try harder to find worktree from gitdir
  2026-05-23 14:06                       ` Johannes Sixt
@ 2026-05-23 15:33                         ` Mark Levedahl
  0 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-23 15:33 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: egg_mushroomcow, bootaina702, git

On 5/23/26 10:06 AM, Johannes Sixt wrote:
> I tried to come up with a situation where we end up here, but couldn't.
> When would this happen? If it actually can't happen, I would prefer to
> spawn fewer git processes and just take the result of 'file dirname'.
>
> If the code must remain, can we please rename one of gitdir_parent or
> parent_gitdir?
There are some odd cases I've seen, mostly driven by network file systems (using Samba or
NFS or ...) that don't behave as POSIX. Perhaps that is reasonably out of scope. The more
I think about this, I'll delete that part.

Symlinks can create odd cases:

Consider /tmp/main  and  /tmp/worktree. The latter has a .git entry that is a symlink to
/tmp/main.git. git rev-parse --absolute-git-dir shows
    in dir /tmp/worktree/.git           /tmp/main.git
So, git-gui would start in /tmp/main, and not in /tmp/worktree.

If using git new-workdir (which is still in the wild), this creates a .git dir with
symlinks to all the subdirs. Now git rev-parse reports
   in dir /tmp/worktree.git          /tmp/worktree/.git
   in dir /tmp/worktree/.git/logs   /tmp/main/.git
So, as long as this is started in the top level of .git, it's ok.

While the shell understands we descended into a symlink and reports pwd does not
de-referencing the symlink, tcl always dereferences the symlinks. So, any ability to
contain this behavior is very likely system and shell dependent.

So, in summary, I'll probably simplify the check to just --show-toplevel run in the dir
above .git.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v2 08/11] git-gui: use HEAD as current branch when detached (bug fix)
  2026-05-20 20:23                   ` [PATCH v2 " Mark Levedahl
                                       ` (6 preceding siblings ...)
  2026-05-20 20:24                     ` [PATCH v2 07/11] git-gui: try harder to find worktree from gitdir Mark Levedahl
@ 2026-05-20 20:24                     ` Mark Levedahl
  2026-05-23 14:08                       ` Johannes Sixt
  2026-05-20 20:24                     ` [PATCH v2 09/11] git-gui: allow specifying path '.' to the browser Mark Levedahl
                                       ` (4 subsequent siblings)
  12 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-20 20:24 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

commit f87a36b697 ("git-gui: use git-branch --show-current", 2024-02-12)
changed git-gui to use git-branch to access refs, rather than directly
reading files as doing the latter is not compatible with the reftable
backend. git branch --show-current reports an empty branch name when the
head is detached, and in this case load_current_branch needs to report
HEAD using special case logic as it did prior to the above commit. Make
it do so.

This addresses an issue with git-gui browser failing with a detached
head.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/git-gui.sh b/git-gui.sh
index aeb7ed3548..a72d8a59ec 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -648,6 +648,9 @@ proc load_current_branch {} {
 
 	set current_branch [git branch --show-current]
 	set is_detached [expr [string length $current_branch] == 0]
+	if {$is_detached} {
+		set current_branch {HEAD}
+	}
 }
 
 auto_load tk_optionMenu
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 08/11] git-gui: use HEAD as current branch when detached (bug fix)
  2026-05-20 20:24                     ` [PATCH v2 08/11] git-gui: use HEAD as current branch when detached (bug fix) Mark Levedahl
@ 2026-05-23 14:08                       ` Johannes Sixt
  0 siblings, 0 replies; 134+ messages in thread
From: Johannes Sixt @ 2026-05-23 14:08 UTC (permalink / raw)
  To: Mark Levedahl, git; +Cc: egg_mushroomcow, bootaina702

Am 20.05.26 um 22:24 schrieb Mark Levedahl:
> commit f87a36b697 ("git-gui: use git-branch --show-current", 2024-02-12)
> changed git-gui to use git-branch to access refs, rather than directly
> reading files as doing the latter is not compatible with the reftable
> backend. git branch --show-current reports an empty branch name when the
> head is detached, and in this case load_current_branch needs to report
> HEAD using special case logic as it did prior to the above commit. Make
> it do so.
> 
> This addresses an issue with git-gui browser failing with a detached
> head.

Nice catch. I'll reorder this as the first commit.

> 
> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
> ---
>  git-gui.sh | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/git-gui.sh b/git-gui.sh
> index aeb7ed3548..a72d8a59ec 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -648,6 +648,9 @@ proc load_current_branch {} {
>  
>  	set current_branch [git branch --show-current]
>  	set is_detached [expr [string length $current_branch] == 0]
> +	if {$is_detached} {
> +		set current_branch {HEAD}
> +	}
>  }
>  
>  auto_load tk_optionMenu

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v2 09/11] git-gui: allow specifying path '.' to the browser
  2026-05-20 20:23                   ` [PATCH v2 " Mark Levedahl
                                       ` (7 preceding siblings ...)
  2026-05-20 20:24                     ` [PATCH v2 08/11] git-gui: use HEAD as current branch when detached (bug fix) Mark Levedahl
@ 2026-05-20 20:24                     ` Mark Levedahl
  2026-05-23 14:23                       ` Johannes Sixt
  2026-05-20 20:24                     ` [PATCH v2 10/11] git-gui: adapt blame/browser parsing for bare operation Mark Levedahl
                                       ` (3 subsequent siblings)
  12 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-20 20:24 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

Invoking "git-gui browser rev ." should show the file browser for the
commitish rev, starting at the current directory. When the current
directory is the working tree root, this errors out in normalize_relpath
because the '.' is removed, yielding an empty list as argument to [file
join ...]. The browser function demands "./" in this case, so make it
so. (./ works on Windows as well because g4w accepts posix file
naming).

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/git-gui.sh b/git-gui.sh
index a72d8a59ec..d373457901 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -3007,7 +3007,11 @@ proc normalize_relpath {path} {
 		}
 		lappend elements $item
 	}
-	return [eval file join $elements]
+	if {$elements ne {}} {
+		return [eval file join $elements]
+	} else {
+		return {./}
+	}
 }
 
 # -- Not a normal commit type invocation?  Do that instead!
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 09/11] git-gui: allow specifying path '.' to the browser
  2026-05-20 20:24                     ` [PATCH v2 09/11] git-gui: allow specifying path '.' to the browser Mark Levedahl
@ 2026-05-23 14:23                       ` Johannes Sixt
  2026-05-23 15:43                         ` Mark Levedahl
  0 siblings, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-23 14:23 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 20.05.26 um 22:24 schrieb Mark Levedahl:
> Invoking "git-gui browser rev ." should show the file browser for the
> commitish rev, starting at the current directory. When the current
> directory is the working tree root, this errors out in normalize_relpath
> because the '.' is removed, yielding an empty list as argument to [file
> join ...]. The browser function demands "./" in this case, so make it
> so. (./ works on Windows as well because g4w accepts posix file
> naming).

I wonder why we need "./" instead of plain ".". The latter works just
fine in my tests (on Linux).

> 
> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
> ---
>  git-gui.sh | 6 +++++-
>  1 file changed, 5 insertions(+), 1 deletion(-)
> 
> diff --git a/git-gui.sh b/git-gui.sh
> index a72d8a59ec..d373457901 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -3007,7 +3007,11 @@ proc normalize_relpath {path} {
>  		}
>  		lappend elements $item
>  	}
> -	return [eval file join $elements]
> +	if {$elements ne {}} {
> +		return [eval file join $elements]
> +	} else {
> +		return {./}
> +	}
>  }
>  
>  # -- Not a normal commit type invocation?  Do that instead!

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 09/11] git-gui: allow specifying path '.' to the browser
  2026-05-23 14:23                       ` Johannes Sixt
@ 2026-05-23 15:43                         ` Mark Levedahl
  0 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-23 15:43 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: egg_mushroomcow, bootaina702, git



On 5/23/26 10:23 AM, Johannes Sixt wrote:
> Am 20.05.26 um 22:24 schrieb Mark Levedahl:
>> Invoking "git-gui browser rev ." should show the file browser for the
>> commitish rev, starting at the current directory. When the current
>> directory is the working tree root, this errors out in normalize_relpath
>> because the '.' is removed, yielding an empty list as argument to [file
>> join ...]. The browser function demands "./" in this case, so make it
>> so. (./ works on Windows as well because g4w accepts posix file
>> naming).
> I wonder why we need "./" instead of plain ".". The latter works just
> fine in my tests (on Linux).
'.' caused errors in browser::new in for me before while './' worked, but now I find both
work. I'm confused, this must have been an interaction with something else in flight at
the time, will revert to '.' if that passes my tests on Windows as well as it is more
consistent of not adding '/' to a dirname.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v2 10/11] git-gui: adapt blame/browser parsing for bare operation
  2026-05-20 20:23                   ` [PATCH v2 " Mark Levedahl
                                       ` (8 preceding siblings ...)
  2026-05-20 20:24                     ` [PATCH v2 09/11] git-gui: allow specifying path '.' to the browser Mark Levedahl
@ 2026-05-20 20:24                     ` Mark Levedahl
  2026-05-21  5:02                       ` Shroom Moo
                                         ` (2 more replies)
  2026-05-20 20:24                     ` [PATCH v2 11/11] git-gui: add gui and pick as explicit subcommands Mark Levedahl
                                       ` (2 subsequent siblings)
  12 siblings, 3 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-20 20:24 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

git-gui's blame and browser subcommands do not work with bare
repositories, but they should per commit c52c94524b ("git-gui: Allow
blame/browser subcommands on bare repositories", 2007-07-17). Assuming
that commit worked, something changed since reintroducing a hard-coded
dependency upon a worktree.

The basic issue goes back to 3e45ee1ef2 ("git-gui: Smarter command line
parsing for browser, blame", 2007-05-08), which seeks to implement
command line parsing similar to git blame. That commit introduces
depencies upon the worktree to decide which argument is rev or path.

Looking at builtin/blame.c in git around line 1120:

	 * (1) if dashdash_pos != 0, it is either
	 *     "blame [revisions] -- <path>" or
	 *     "blame -- <path> <rev>"
	 *
	 * (2) otherwise, it is one of the two:
	 *     "blame [revisions] <path>"
	 *     "blame <path> <rev>"

shows the clear intent: rev and path may be swapped in input so both
meanings must be tried, but -- may be used to designate which is the
path forcing or precluding trying the swapped arguments.

With a worktree, git gui correctly swaps the arguments if the given path
exists in the worktree. git blame does this using the git repository.
But, git-gui sometimes interprets the -- to have an exactly opposite
meaning:

    git blame       Makefile gitgui-0.19.0       works
    git gui blame   Makefile gitgui-0.19.0       works

    git blame       -- Makefile gitgui-0.19.0    works
    git gui blame   -- Makefile gitgui-0.19.0    works

    git blame       Makefile -- gitgui-0.19.0    fails (correctly)
    git gui blame   Makefile -- gitgui-0.19.0    works (should fail)

    git blame       gitgui-0.19.0 -- Makefile    works (correctly)
    git gui blame   gitgui-0.19.0 -- Makefile    fails (should work)

It is possible to patch the code to operate without a worktree, but this
will make the commands operate differently with and without a worktree,
won't fix the parsing issues above, and won't address the issues that
can arise when using a worktree to help decisions on a different rev
with file/directory conflicts, etc.

So, let's rework the parser so that it uses -- as does git blame, and
uses git ls-tree to query the given revision for existence and type of
path rather than basing this upon a possibly unrelated worktree. Also,
abort early when the given path is not found, or does not match the need
(file or directory). This fixes some current cases where git-gui will
open a window with no content, possibly also with an error message.

This does not change whether or how git-gui uses staged and unstaged
content in the current worktree for blame display.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 151 ++++++++++++++++++++++++++++++-----------------------
 1 file changed, 87 insertions(+), 64 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index d373457901..ae609f86f1 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -3014,100 +3014,123 @@ proc normalize_relpath {path} {
 	}
 }
 
+proc find_path_type {head path} {
+	if {$path eq {./}} {
+		# the root-tree exists in every rev, ls-tree gives data on the contents,
+		# not the type of tree itself. So, if the rev exists, return {tree}
+		if {[catch {set objtype [git ls-tree $head]}]} {
+			set objtype {}
+		} else {
+			set objtype {tree}
+		}
+	} else {
+		# test that the path exists in head, ls-tree gives info on the path only
+		if {[catch {set objtype [git ls-tree {--format=%(objecttype)} $head $path]}]} {
+			set objtype {}
+		}
+	}
+	return $objtype
+}
+
 # -- Not a normal commit type invocation?  Do that instead!
 #
 switch -- $subcommand {
 browser -
 blame {
 	if {$subcommand eq "blame"} {
-		set subcommand_args {[--line=<num>] rev? path}
+		set subcommand_args {[--line=<num>] <[rev] [--] filename | [--] filename rev>}
+		set required_objtype blob
 	} else {
-		set subcommand_args {rev? path}
+		set subcommand_args {<[rev] [--] directory | [--] directory rev>}
+		set required_objtype tree
 	}
-	if {$argv eq {}} usage
+	set maxargs [llength $subcommand_args]
+	set nargs [llength $argv]
+	if {$nargs < 1 || $nargs > $maxargs} usage
 	set head {}
+	set althead {}
 	set path {}
+	set altpath {}
+	set canswap 1
 	set jump_spec {}
-	set is_path 0
-	foreach a $argv {
-		set p [file join $_prefix $a]
 
-		if {$is_path || [file exists $p]} {
-			if {$path ne {}} usage
-			set path [normalize_relpath $p]
-			break
-		} elseif {$a eq {--}} {
-			if {$path ne {}} {
-				if {$head ne {}} usage
-				set head $path
-				set path {}
+	# assume: [--line=num] [head] [--] path as the possible arguments, in order.
+	# head and path may need a swap later.
+	for {set iarg 0} {$iarg < $nargs} {incr iarg} {
+		set arg [lindex $argv $iarg]
+		if {$arg eq {--}} {
+			# next arg is the path, prevent or FORCE swap?
+			if {$iarg == $nargs - 2} {
+				set canswap 0
+			} elseif {$iarg == $nargs - 3} {
+				set canswap 2
+			} else {
+				usage
 			}
-			set is_path 1
-		} elseif {[regexp {^--line=(\d+)$} $a a lnum]} {
-			if {$jump_spec ne {} || $head ne {}} usage
+		} elseif {[regexp {^--line=(\d+)$} $arg arg lnum]} {
+			# --line can only be the first arg
+			if {$iarg != 0 || $maxargs < 4} usage
 			set jump_spec [list $lnum]
+		} elseif {$iarg == $nargs - 1} {
+			# assume final argument is path
+			set path [normalize_relpath [file join $_prefix $arg]]
+			set althead $arg
 		} elseif {$head eq {}} {
-			if {$head ne {}} usage
-			set head $a
-			set is_path 1
+			# assume the other argument is head
+			set head $arg
+			set altpath [normalize_relpath [file join $_prefix $arg]]
 		} else {
 			usage
 		}
 	}
-	unset is_path
-
-	if {$head ne {} && $path eq {}} {
-		if {[string index $head 0] eq {/}} {
-			set path [normalize_relpath $head]
-			set head {}
-		} else {
-			set path [normalize_relpath $_prefix$head]
-			set head {}
-		}
-	}
 
+	# no swapping allowed if head not given, use current branch (HEAD)
 	if {$head eq {}} {
 		load_current_branch
-	} else {
-		if {[regexp [string map "@@ [expr $hashlength - 1]" {^[0-9a-f]{1,@@}$}] $head]} {
-			if {[catch {
-					set head [git rev-parse --verify $head]
-				} err]} {
-				if {[tk windowingsystem] eq "win32"} {
-					tk_messageBox -icon error -title [mc Error] -message $err
-				} else {
-					puts stderr $err
-				}
-				exit 1
-			}
+		set head $current_branch
+		set canswap 0
+	}
+
+	# -- before "rev" arg means we got -- path head
+	if {$canswap == 2} {
+		set head $althead
+		set path $altpath
+		set canswap 0
+	}
+
+	set objtype [find_path_type $head $path]
+	if {$objtype eq {} && $canswap} {
+		set objtype [find_path_type $althead $altpath]
+		if {$objtype ne {}} {
+			set head $althead
+			set path $altpath
 		}
-		set current_branch $head
+	}
+	set current_branch $head
+
+	# check that path exists in head, and objtype matches need
+	if {$objtype ne $required_objtype} {
+		switch -- $required_objtype {
+			tree {set err [strcat \
+				[mc "'%s' is not a directory in rev '%s'" $path $head]]}
+			blob {set err [strcat \
+				[mc "'%s' is not a filename in rev '%s'" $path $head]]}
+		}
+		if {[tk windowingsystem] eq "win32"} {
+			catch {wm withdraw .}
+			error_popup $err
+		} else {
+			puts stderr $err
+		}
+		exit 1
 	}
 
 	wm deiconify .
 	switch -- $subcommand {
 	browser {
-		if {$jump_spec ne {}} usage
-		if {$head eq {}} {
-			if {$path ne {} && [file isdirectory $path]} {
-				set head $current_branch
-			} else {
-				set head $path
-				set path {}
-			}
-		}
 		browser::new $head $path
 	}
-	blame   {
-		if {$head eq {} && ![file exists $path]} {
-			catch {wm withdraw .}
-			tk_messageBox \
-				-icon error \
-				-type ok \
-				-title [mc "git-gui: fatal error"] \
-				-message [mc "fatal: cannot stat path %s: No such file or directory" $path]
-			exit 1
-		}
+	blame {
 		blame::new $head $path $jump_spec
 	}
 	}
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 10/11] git-gui: adapt blame/browser parsing for bare operation
  2026-05-20 20:24                     ` [PATCH v2 10/11] git-gui: adapt blame/browser parsing for bare operation Mark Levedahl
@ 2026-05-21  5:02                       ` Shroom Moo
  2026-05-21 17:35                         ` Mark Levedahl
  2026-05-23 18:19                       ` [PATCH] fixup git-gui: allow blame to show uncommitted changes Mark Levedahl
  2026-05-23 20:31                       ` [PATCH v2 10/11] git-gui: adapt blame/browser parsing for bare operation Johannes Sixt
  2 siblings, 1 reply; 134+ messages in thread
From: Shroom Moo @ 2026-05-21  5:02 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: git, Johannes Sixt, Aina Boot

On 5/21/26 4:24 AM, Mark Levedahl wrote: 
> +proc find_path_type {head path} {
> +	if {$path eq {./}} {
> +		# the root-tree exists in every rev, ls-tree gives data on the contents,
> +		# not the type of tree itself. So, if the rev exists, return {tree}
> +		if {[catch {set objtype [git ls-tree $head]}]} {
> +			set objtype {}
> +		} else {
> +			set objtype {tree}
> +		}
> +	} else {
> +		# test that the path exists in head, ls-tree gives info on the path only
> +		if {[catch {set objtype [git ls-tree {--format=%(objecttype)} $head $path]}]} {
> +			set objtype {}
> +		}
> +	}
> +	return $objtype
> +}

In v1, argument parsing relied on file exists within the worktree to 
determine if a path existed, without using ls-tree. In v2, the use of 
git ls-tree seems to actually be intended to list directory contents, 
rather than querying the type of the path itself. 

If $path is a directory (a tree object), git ls-tree outputs the 
object type for every entry within that directory, one per line. 

The variable objtype is assigned a multi-line string. When compared 
against "tree", the match fails, causing the function to return an 
empty string, which subsequently leads to an error. We can change to 
"git cat-file -t" or similiar approaches. 

Shroom

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 10/11] git-gui: adapt blame/browser parsing for bare operation
  2026-05-21  5:02                       ` Shroom Moo
@ 2026-05-21 17:35                         ` Mark Levedahl
  2026-05-22 15:05                           ` Shroom Moo
  0 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-21 17:35 UTC (permalink / raw)
  To: Shroom Moo; +Cc: git, Johannes Sixt, Aina Boot



On 5/21/26 1:02 AM, Shroom Moo wrote:
> On 5/21/26 4:24 AM, Mark Levedahl wrote: 
>> +proc find_path_type {head path} {
>> +	if {$path eq {./}} {
>> +		# the root-tree exists in every rev, ls-tree gives data on the contents,
>> +		# not the type of tree itself. So, if the rev exists, return {tree}
>> +		if {[catch {set objtype [git ls-tree $head]}]} {
>> +			set objtype {}
>> +		} else {
>> +			set objtype {tree}
>> +		}
>> +	} else {
>> +		# test that the path exists in head, ls-tree gives info on the path only
>> +		if {[catch {set objtype [git ls-tree {--format=%(objecttype)} $head $path]}]} {
>> +			set objtype {}
>> +		}
>> +	}
>> +	return $objtype
>> +}
> In v1, argument parsing relied on file exists within the worktree to 
> determine if a path existed, without using ls-tree. In v2, the use of 
> git ls-tree seems to actually be intended to list directory contents, 
> rather than querying the type of the path itself. 
>
> If $path is a directory (a tree object), git ls-tree outputs the 
> object type for every entry within that directory, one per line. 
>
> The variable objtype is assigned a multi-line string. When compared 
> against "tree", the match fails, causing the function to return an 
> empty string, which subsequently leads to an error. We can change to 
> "git cat-file -t" or similiar approaches. 
>
> Shroom
>
git ls-tree $rev $path --format='%(objecttype)' gives the type of the object at $path in
$rev, or an error. The types returned are "tree" for a directory, "blob" for a file. So
this gives definitive information if the object desired exists in the given rev, and is of
the right type. (We don't care about commits and tags, those cannot be blamed or browsed).

git-lstree $rev $dirname/  lists all of the objects in the directory, while git-lstree
$rev $dirname (no trailing /) gives info on the directory itself. There is no name for the
root directory itself, all of its contents are listed. That is why the root './' is
special case.

Asking the worktree that is on version 33 about whether frotz is a directory in version 2
is just asking for trouble, at best the worktree is authoritative for the checked out
version, but even then there can be uncommitted changed. In the root of git-gui, I get

/git-gui.sh browser gitgui-0.9.0 Makefile
'Makefile' is not a directory in rev 'gitgui-0.9.0'

so the types of objects are being checked.

Mark


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 10/11] git-gui: adapt blame/browser parsing for bare operation
  2026-05-21 17:35                         ` Mark Levedahl
@ 2026-05-22 15:05                           ` Shroom Moo
  0 siblings, 0 replies; 134+ messages in thread
From: Shroom Moo @ 2026-05-22 15:05 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: git, Johannes Sixt, Aina Boot

On 5/22/26 1:35 AM, Mark Levedahl wrote: 
> Asking the worktree that is on version 33 about whether frotz is a directory in version 2
> is just asking for trouble, at best the worktree is authoritative for the checked out
> version, but even then there can be uncommitted changed. In the root of git-gui, I get
> 
> /git-gui.sh browser gitgui-0.9.0 Makefile
> 'Makefile' is not a directory in rev 'gitgui-0.9.0'
> 
> so the types of objects are being checked.
> 
> Mark

You are right. The input is subdir instead of subdir/. All the 
current tests work well. Patch 10/11 does not cause damage to the 
functionality of browser and blame commands. 

Shroom


^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH] fixup git-gui: allow blame to show uncommitted changes
  2026-05-20 20:24                     ` [PATCH v2 10/11] git-gui: adapt blame/browser parsing for bare operation Mark Levedahl
  2026-05-21  5:02                       ` Shroom Moo
@ 2026-05-23 18:19                       ` Mark Levedahl
  2026-05-23 20:31                       ` [PATCH v2 10/11] git-gui: adapt blame/browser parsing for bare operation Johannes Sixt
  2 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-23 18:19 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

Commit a0db0d61fb ("git-gui: Generate blame on uncommitted working tree
file", 2007-05-08) added ability for git-gui's blame to use uncommited
content in the worktree in the blame display. The specific mechanism
that allows this is passing head as {} to the blame constructor, and this
mode is enabled by specifying no head, which means the current branch is
used, and no swapping of head / path is possible. Path checking looks for the
path existing in the currently checked out branch, so files unknown to
git cannot be blamed. This mirrors git blame behavior. Both now will

    use the worktree contents of the file $path as the basis for blame,
        including showing an empty blame if the file exists and is empty.
    error if $path is remove from the worktree, even if it exists on the
          current branch.

The latter behavior is a change: before this patch, git-gui will show
the unknown file in its entirety with no annotations. The new behavior,
following git blame, reports that git knows nothing about the file.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
This patch will be squashed into 0011 in a v3, or kept separate. The
comment in patch 11 saying blame's use of the worktree is unaffected
is wrong, this fixes blame to do exactly what git-blame does with
worktree content. Slightly different than what git-gui did before
regarding unknown files, but I think git-blame's approach is correct.


 git-gui.sh | 29 ++++++++++++++++++++++++-----
 1 file changed, 24 insertions(+), 5 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index ae609f86f1..114511974a 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -3084,11 +3084,16 @@ blame {
 		}
 	}
 
-	# no swapping allowed if head not given, use current branch (HEAD)
+	# If head not given, use current branch (HEAD), no swapping allowed,
+	# and blame may use the worktree file content.
+	set use_worktree 0
 	if {$head eq {}} {
 		load_current_branch
 		set head $current_branch
 		set canswap 0
+		if {$subcommand eq {blame} && ![is_bare]} {
+			set use_worktree 1
+		}
 	}
 
 	# -- before "rev" arg means we got -- path head
@@ -3098,7 +3103,16 @@ blame {
 		set canswap 0
 	}
 
-	set objtype [find_path_type $head $path]
+	if {$use_worktree} {
+		if {[file isfile $path]} {
+			set objtype {blob}
+		} else {
+			set objtype {}
+		}
+	} else {
+		set objtype [find_path_type $head $path]
+	}
+
 	if {$objtype eq {} && $canswap} {
 		set objtype [find_path_type $althead $altpath]
 		if {$objtype ne {}} {
@@ -3108,7 +3122,7 @@ blame {
 	}
 	set current_branch $head
 
-	# check that path exists in head, and objtype matches need
+	# check objtype matches need
 	if {$objtype ne $required_objtype} {
 		switch -- $required_objtype {
 			tree {set err [strcat \
@@ -3130,8 +3144,13 @@ blame {
 	browser {
 		browser::new $head $path
 	}
-	blame {
-		blame::new $head $path $jump_spec
+
+	blame   {
+		if {$use_worktree} {
+			blame::new {} $path $jump_spec
+		} else {
+			blame::new $head $path $jump_spec
+		}
 	}
 	}
 	return
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 10/11] git-gui: adapt blame/browser parsing for bare operation
  2026-05-20 20:24                     ` [PATCH v2 10/11] git-gui: adapt blame/browser parsing for bare operation Mark Levedahl
  2026-05-21  5:02                       ` Shroom Moo
  2026-05-23 18:19                       ` [PATCH] fixup git-gui: allow blame to show uncommitted changes Mark Levedahl
@ 2026-05-23 20:31                       ` Johannes Sixt
  2 siblings, 0 replies; 134+ messages in thread
From: Johannes Sixt @ 2026-05-23 20:31 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 20.05.26 um 22:24 schrieb Mark Levedahl:
> git-gui's blame and browser subcommands do not work with bare
> repositories, but they should per commit c52c94524b ("git-gui: Allow
> blame/browser subcommands on bare repositories", 2007-07-17). Assuming
> that commit worked, something changed since reintroducing a hard-coded
> dependency upon a worktree.
> 
> The basic issue goes back to 3e45ee1ef2 ("git-gui: Smarter command line
> parsing for browser, blame", 2007-05-08), which seeks to implement
> command line parsing similar to git blame. That commit introduces
> depencies upon the worktree to decide which argument is rev or path.
> 
> Looking at builtin/blame.c in git around line 1120:
> 
> 	 * (1) if dashdash_pos != 0, it is either
> 	 *     "blame [revisions] -- <path>" or
> 	 *     "blame -- <path> <rev>"
> 	 *
> 	 * (2) otherwise, it is one of the two:
> 	 *     "blame [revisions] <path>"
> 	 *     "blame <path> <rev>"
> 
> shows the clear intent: rev and path may be swapped in input so both
> meanings must be tried, but -- may be used to designate which is the
> path forcing or precluding trying the swapped arguments.

Please do not use this code comment as recipe for our own argument
parseing. In particular, that <path> can occur for <rev> goes back to
the initial implementation of git pickaxe in cee7f245dcae ("git-pickaxe:
blame rewritten.", 2006-10-19). Since acca687fa9db ("git-pickaxe: retire
pickaxe", 2006-11-08), the documentation of git-blame states that <file>
is always last (but the implementation was not adjusted accordingly).

In general, Git's argument parsing requires revisions before pathspec.
To disambiguate, '--' can be used. If it is not used, arguments are
check whether they are files or refs, and as soon as one argument is
identified as file unambiguously, all later arguments must also be files.

We should follow this pattern, and to do that, we could just delegate
argument processing to `git rev-parse`.

> 
> With a worktree, git gui correctly swaps the arguments if the given path
> exists in the worktree. git blame does this using the git repository.
> But, git-gui sometimes interprets the -- to have an exactly opposite
> meaning:
> 
>     git blame       Makefile gitgui-0.19.0       works
>     git gui blame   Makefile gitgui-0.19.0       works

Git gui shows something, but ignores the ref, so doesn't quite work.

> 
>     git blame       -- Makefile gitgui-0.19.0    works
>     git gui blame   -- Makefile gitgui-0.19.0    works

Ditto.

> 
>     git blame       Makefile -- gitgui-0.19.0    fails (correctly)
>     git gui blame   Makefile -- gitgui-0.19.0    works (should fail)

Ditto.

> 
>     git blame       gitgui-0.19.0 -- Makefile    works (correctly)
>     git gui blame   gitgui-0.19.0 -- Makefile    fails (should work)

Yes, there's a bug in the argument parser that -- isn't skipped, but
treated as the file name.

> 
> It is possible to patch the code to operate without a worktree, but this
> will make the commands operate differently with and without a worktree,
> won't fix the parsing issues above, and won't address the issues that
> can arise when using a worktree to help decisions on a different rev
> with file/directory conflicts, etc.

Before this patch 'git gui blame' can show contents uncommitted changes,
but with this patch this isn't possible. I see you have just sent a
patch that may fix this, but I havn't looked at it, yet.

> 
> So, let's rework the parser so that it uses -- as does git blame, and
> uses git ls-tree to query the given revision for existence and type of
> path rather than basing this upon a possibly unrelated worktree. Also,
> abort early when the given path is not found, or does not match the need
> (file or directory). This fixes some current cases where git-gui will
> open a window with no content, possibly also with an error message.

There is no desire to make 'git gui blame' work the same with and
without a working tree.

If we invoke git to help argument parsing, then it should be
'rev-parse', not 'ls-tree'.

> 
> This does not change whether or how git-gui uses staged and unstaged
> content in the current worktree for blame display.
> 
> Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
> ---

>  	wm deiconify .
>  	switch -- $subcommand {
>  	browser {
> -		if {$jump_spec ne {}} usage

Let's keep this line, which diagnoses an incorrect --line= argument.

> -		if {$head eq {}} {
> -			if {$path ne {} && [file isdirectory $path]} {
> -				set head $current_branch
> -			} else {
> -				set head $path
> -				set path {}
> -			}
> -		}
>  		browser::new $head $path
>  	}
> -	blame   {
> -		if {$head eq {} && ![file exists $path]} {
> -			catch {wm withdraw .}
> -			tk_messageBox \
> -				-icon error \
> -				-type ok \
> -				-title [mc "git-gui: fatal error"] \
> -				-message [mc "fatal: cannot stat path %s: No such file or directory" $path]
> -			exit 1
> -		}
> +	blame {
>  		blame::new $head $path $jump_spec
>  	}
>  	}

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v2 11/11] git-gui: add gui and pick as explicit subcommands
  2026-05-20 20:23                   ` [PATCH v2 " Mark Levedahl
                                       ` (9 preceding siblings ...)
  2026-05-20 20:24                     ` [PATCH v2 10/11] git-gui: adapt blame/browser parsing for bare operation Mark Levedahl
@ 2026-05-20 20:24                     ` Mark Levedahl
  2026-05-24  7:00                       ` Johannes Sixt
  2026-05-24  7:16                     ` [PATCH v2 00/11] Improve git gui operation without a worktree Johannes Sixt
  2026-05-31 23:02                     ` [PATCH v3 00/12] " Mark Levedahl
  12 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-20 20:24 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

git-gui accepts subcommands blame | browser | citool, and assumes the
subcommand is 'gui' if none is actually given, But, git-gui also has a
repository picker (choose_repository::pick) that can create a new
repository + worktree, or choose an existing one, switch to that, and
the run the gui. The user has no direct control over invoking the
picker, instead the picker is triggered by failure in the repository /
worktree discovery process: this includes being started in a directory
not controlled by git, which is probably the intended use case.

The picker can appear when the user has no intention of creating a new
worktree, and the user cannot use the picker to create a new worktree
inside another.

So, add two explicit subcommands:
    gui  - Run the gui if repository/worktree discovery succeeds, or die
           with an error message, but never run the picker.
    pick - First run the picker, regardless, then start the gui in
           the chosen worktree.

Nothing in this changes the prior behavior, the alternates above must be
explicitly selected to see any change.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 25 +++++++++++++++++++++----
 1 file changed, 21 insertions(+), 4 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index ae609f86f1..299c1a0292 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1024,6 +1024,8 @@ proc load_config {include_global} {
 ##
 ## feature option selection
 
+enable_option picker
+enable_option gitdir_discovery
 if {[regexp {^git-(.+)$} [file tail $argv0] _junk subcommand]} {
 	unset _junk
 } else {
@@ -1035,6 +1037,9 @@ if {$subcommand eq {gui.sh}} {
 if {$subcommand eq {gui} && [llength $argv] > 0} {
 	set subcommand [lindex $argv 0]
 	set argv [lrange $argv 1 end]
+	if {$subcommand eq {gui}} {
+		disable_option picker
+	}
 }
 
 enable_option multicommit
@@ -1050,6 +1055,7 @@ blame {
 	disable_option multicommit
 	disable_option branch
 	disable_option transport
+	disable_option picker
 }
 citool {
 	enable_option singlecommit
@@ -1058,6 +1064,7 @@ citool {
 	disable_option multicommit
 	disable_option branch
 	disable_option transport
+	disable_option picker
 
 	while {[llength $argv] > 0} {
 		set a [lindex $argv 0]
@@ -1080,6 +1087,9 @@ citool {
 		set argv [lrange $argv 1 end]
 	}
 }
+pick {
+	disable_option gitdir_discovery
+}
 }
 
 ######################################################################
@@ -1174,7 +1184,7 @@ proc unset_gitdir_vars {} {
 
 # find repository.
 set _gitdir {}
-if {$_gitdir eq {}} {
+if {[is_enabled gitdir_discovery]} {
 	if {[catch {
 		set _gitdir [git rev-parse --absolute-git-dir]
 	} err]} {
@@ -1186,7 +1196,7 @@ if {$_gitdir eq {}} {
 }
 
 set picked 0
-if {$_gitdir eq {}} {
+if {$_gitdir eq {} && [is_enabled picker]} {
 	unset_gitdir_vars
 	load_config 1
 	apply_config
@@ -1202,6 +1212,12 @@ if {$_gitdir eq {}} {
 	set picked 1
 }
 
+if {$_gitdir eq {}} {
+	catch {wm withdraw .}
+	error_popup [strcat [mc "Git directory not found:"] "\n\n$err"]
+	exit 1
+}
+
 # find worktree, continue without if not required
 if {[catch {
 	set _gitworktree [git rev-parse --show-toplevel]
@@ -3137,14 +3153,15 @@ blame {
 	return
 }
 citool -
-gui {
+gui -
+pick {
 	if {[llength $argv] != 0} {
 		usage
 	}
 	# fall through to setup UI for commits
 }
 default {
-	set err "[mc usage:] $argv0 \[{blame|browser|citool}\]"
+	set err "[mc usage:] $argv0 \[{blame|browser|citool|gui|pick}\]"
 	if {[tk windowingsystem] eq "win32"} {
 		wm withdraw .
 		tk_messageBox -icon error -message $err \
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 11/11] git-gui: add gui and pick as explicit subcommands
  2026-05-20 20:24                     ` [PATCH v2 11/11] git-gui: add gui and pick as explicit subcommands Mark Levedahl
@ 2026-05-24  7:00                       ` Johannes Sixt
  0 siblings, 0 replies; 134+ messages in thread
From: Johannes Sixt @ 2026-05-24  7:00 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 20.05.26 um 22:24 schrieb Mark Levedahl:
> git-gui accepts subcommands blame | browser | citool, and assumes the
> subcommand is 'gui' if none is actually given, But, git-gui also has a
> repository picker (choose_repository::pick) that can create a new
> repository + worktree, or choose an existing one, switch to that, and
> the run the gui. The user has no direct control over invoking the
> picker, instead the picker is triggered by failure in the repository /
> worktree discovery process: this includes being started in a directory
> not controlled by git, which is probably the intended use case.
> 
> The picker can appear when the user has no intention of creating a new
> worktree, and the user cannot use the picker to create a new worktree
> inside another.
> 
> So, add two explicit subcommands:
>     gui  - Run the gui if repository/worktree discovery succeeds, or die
>            with an error message, but never run the picker.
>     pick - First run the picker, regardless, then start the gui in
>            the chosen worktree.
> 
> Nothing in this changes the prior behavior, the alternates above must be
> explicitly selected to see any change.

Good.

> @@ -1174,7 +1184,7 @@ proc unset_gitdir_vars {} {
>  
>  # find repository.
>  set _gitdir {}
> -if {$_gitdir eq {}} {
> +if {[is_enabled gitdir_discovery]} {

This makes a factually unconditional branch into a conditional one.

>  	if {[catch {
>  		set _gitdir [git rev-parse --absolute-git-dir]
>  	} err]} {
> @@ -1186,7 +1196,7 @@ if {$_gitdir eq {}} {
>  }
>  
>  set picked 0
> -if {$_gitdir eq {}} {
> +if {$_gitdir eq {} && [is_enabled picker]} {
>  	unset_gitdir_vars
>  	load_config 1
>  	apply_config
> @@ -1202,6 +1212,12 @@ if {$_gitdir eq {}} {
>  	set picked 1
>  }
>  
> +if {$_gitdir eq {}} {
> +	catch {wm withdraw .}
> +	error_popup [strcat [mc "Git directory not found:"] "\n\n$err"]

I wondered where this $err is filled in, and it can only be the error
from a failed gitdir discovery. Good.

> +	exit 1
> +}
> +
>  # find worktree, continue without if not required
>  if {[catch {
>  	set _gitworktree [git rev-parse --show-toplevel]
This looks good!

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 00/11] Improve git gui operation without a worktree
  2026-05-20 20:23                   ` [PATCH v2 " Mark Levedahl
                                       ` (10 preceding siblings ...)
  2026-05-20 20:24                     ` [PATCH v2 11/11] git-gui: add gui and pick as explicit subcommands Mark Levedahl
@ 2026-05-24  7:16                     ` Johannes Sixt
  2026-05-25 16:02                       ` Mark Levedahl
  2026-05-31 23:02                     ` [PATCH v3 00/12] " Mark Levedahl
  12 siblings, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-05-24  7:16 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 20.05.26 um 22:23 schrieb Mark Levedahl:
> git gui has a number of inter-related problems that result in problems
> during startup from anything but a checked out worktree pointing at a
> valid git repository. Some of the symptoms are:
> - blame / browser subcommands, and launching gitk, are intended to be
>   useful without a worktree, but fail to work.
> - unlike git, git-gui is supposed to use the parent directory as a
>   worktree if started from the .git subdirectory in the very common
>   single worktree + embedded git repository format. This does not
>   work.
> - git-gui includes a repository picker allowing a user to select a
>   worktree from a list and/or start a new repo+worktree: this dialog can
>   appear at unexpected times, masking useful error feedback on
>   configuration problems.
> 
> This patch series addresses the above issues, substantially rewriting
> the initial repository/worktree process to rely upon git rev-parse so
> that git's knowledge of access rules, repository configuration, and use
> of GIT_DIR / GIT_WORK_TREE (or git --gitdir / --work-tree) is used
> throughout, replacing code largely based upon what git did in 2008. This
> also means that git gui will naturally gain any new rules implmented in
> git-core.
> 
> With this, git-gui only exports GIT_WORK_TREE when non-empty.
> GIT_WORK_TREE is needed, and must be exported, if the user is overriding
> core.worktree in the git repository. But, GIT_WORK_TREE cannot be used
> to specify the lack of a worktree, so exporting an empty GIT_WORK_TREE
> is one of the problems fixed by this series.
> 
> v2 of this series is a very substantial rewrite driven by j6t's review,
> with patches reoranized and squashed, interfaces to the repository
> chooser changed, a different code structure to allow user control of the
> repository picker, a different approach to fixing the command line
> parser for blame / browser, and other more minor changes. Patches
> for fixing blame / browser are now after all discovery refactoring as
> they cannot be tested without some of those fixes.
> 
> Many subtle things are fixed beyond the list at the top, including
> better compatibility with git blame and repeatable browser / blame
> operation for specific revs not in the worktree, regardless of the
> worktree state. j6t indicated that in the git-gui project, the following
> fails in the current release:
> 
> cd lib
> GIT_DIR=$PWD/../.git GIT_WORK_TREE=$PWD/.. ../git-gui.sh browser origin/master .
> 
> This is due to a _prefix issue, and is fixed as of the patch
>      git-gui: use git rev-parse for worktree discovery
> 

I've completed my review of this iteration.

Repository and working tree discovery is already converging fast.
However, I have issues with the proposed argument parsing of the browser
and blame modes, in particular, I don't think that we need to
accommodate the uncanny file-before-rev argument order and that it
disregards the worktree completely. Maybe we should postpone any changes
in this area, if possible?

Throughout, we use a strange indentation style of 'if {[catch ...' that
is violated in new code, but I left uncommented. It should indent the
catch body one additional level like so:

	if {catch {
			commands that can fail
		} err]} {
		error handling here
	}

Thank you very much for working on this topic.

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v2 00/11] Improve git gui operation without a worktree
  2026-05-24  7:16                     ` [PATCH v2 00/11] Improve git gui operation without a worktree Johannes Sixt
@ 2026-05-25 16:02                       ` Mark Levedahl
  0 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-25 16:02 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: egg_mushroomcow, bootaina702, git

On 5/24/26 3:16 AM, Johannes Sixt wrote:
> Am 20.05.26 um 22:23 schrieb Mark Levedahl:
> I've completed my review of this iteration.
>
> Repository and working tree discovery is already converging fast.
> However, I have issues with the proposed argument parsing of the browser
> and blame modes, in particular, I don't think that we need to
> accommodate the uncanny file-before-rev argument order and that it
> disregards the worktree completely. Maybe we should postpone any changes
> in this area, if possible?
I'm not willing to give up on browser/blame yet: making these work without a worktree was
my motivator to start this series. Ignoring implementation details, etc., the issues I see
here are:

The undocumented feature to accept rev / path or path / rev, as does git blame. The latter
at least has a comment on what is expected in the code, but no mention of this exists in
blames man-page, nor that of any other git command I've examined. I'd prefer to just
remove it, I'll take your comment above as agreement in principal.

browser and blame are both fundamentally about git history. Considering browser.tcl and
blame.tcl, which produce the displays:

browser shows only content from a tree in a commit. It never uses any information from a
worktree. Given a non-existent path (or a non-existent rev),
browser displays an empty window, not an error message.

blame can take content from a file in a commit, or from the worktree as long as the file
is in a commit. A modified file in the worktree has changed / added lines annotated as
uncommitted work. But, given a file not in rev, blame displays the file with no
annotations at all, not as uncommitted work, and no error message.

So. both blame and browser require that $path is contained in $rev, even if $rev defaults
to HEAD.
The parser never checks this, though.

These commands, that *should* work fine without a worktree do not. and display confusing
information (e.g., a blank browser window, or unannotated file) when a simple error
message from the parser would convey more information.

>
> Throughout, we use a strange indentation style of 'if {[catch ...' that
> is violated in new code, but I left uncommented. It should indent the
> catch body one additional level like so:
>
> 	if {catch {
> 			commands that can fail
> 		} err]} {
> 		error handling here
> 	}
Yes, just count the number of { - number of }. Vim's indent mode for tcl gets this very
wrong. All fixed, I hope.
>
> Thank you very much for working on this topic.
and thank you for the very thorough review.
>
> -- Hannes
>

^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v3 00/12] Improve git gui operation without a worktree
  2026-05-20 20:23                   ` [PATCH v2 " Mark Levedahl
                                       ` (11 preceding siblings ...)
  2026-05-24  7:16                     ` [PATCH v2 00/11] Improve git gui operation without a worktree Johannes Sixt
@ 2026-05-31 23:02                     ` Mark Levedahl
  2026-05-31 23:02                       ` [PATCH v3 01/12] git-gui: use HEAD as current branch when detached Mark Levedahl
                                         ` (12 more replies)
  12 siblings, 13 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-31 23:02 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

git gui has a number of inter-related problems that result in problems
during startup from anything but a checked out worktree pointing at a
valid git repository. Some of the symptoms are:
- blame / browser subcommands, and launching gitk, are intended to be
  useful without a worktree, but fail to work.
- unlike git, git-gui is supposed to use the parent directory as a
  worktree if started from the .git subdirectory in the very common
  single worktree + embedded git repository format. This does not
  work.
- git-gui includes a repository picker allowing a user to select a
  worktree from a list and/or start a new repo+worktree: this dialog can
  appear at unexpected times, masking useful error feedback on
  configuration problems.

This patch series addresses the above issues, substantially rewriting
the initial repository/worktree process to rely upon git rev-parse so
that git's knowledge of access rules, repository configuration, and use
of GIT_DIR / GIT_WORK_TREE (or git --gitdir / --work-tree) is used
throughout, replacing code largely based upon what git did in 2008. This
also means that git gui will naturally gain any new rules implmented in
git-core.

With this, git-gui only exports GIT_WORK_TREE when non-empty.
GIT_WORK_TREE is needed, and must be exported, if the user is overriding
core.worktree in the git repository. But, GIT_WORK_TREE cannot be used
to specify the lack of a worktree, so exporting an empty GIT_WORK_TREE
is one of the problems fixed by this series.

v3 of this series addresses j6t's review of v2, with some reordering of
patches (1 from j6t added, patch #8 moved to #1), adds another rewrite
of the browser / blame parser that eliminates the notion of path before
rev on the command line, blame works correctly with a user modified file
in the worktree. Clarification is added on the need for GTI_WORK_TREE,
and the logic in finding a worktree from the gitdir is simplified.

Johannes Sixt (1):
  git-gui: remove unnecessary 'cd $_gitworktree' from do_gitk

Mark Levedahl (11):
  git-gui: use HEAD as current branch when detached
  git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE
  git-gui: do not change global vars in choose_repository::pick
  git-gui: use --absolute-git-dir
  git-gui: use rev-parse exclusively to find a repository
  git-gui: use git rev-parse for worktree discovery
  git-gui: simplify [is_bare] to report if a worktree is known
  git-gui: try harder to find worktree from gitdir
  git-gui: allow specifying path '.' to the browser
  git-gui: check browser/blame arguments carefully
  git-gui: add gui and pick as explicit subcommands

 git-gui.sh                | 377 ++++++++++++++++++++++----------------
 lib/choose_repository.tcl |  21 +--
 2 files changed, 223 insertions(+), 175 deletions(-)

Interdiff against v2:
diff --git a/git-gui.sh b/git-gui.sh
index 299c1a0292..933e72c9b2 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1114,33 +1114,29 @@ unset argv0dir
 ## repository setup
 
 proc find_worktree_from_gitdir {} {
-	# Directory 'parent' of a repository named 'parent/.git' might be the worktree.
-	# Assure parent is a worktree and using the git repository already discovered.
-	# Also, handle case of being in a worktree's gitdir, where file "gitdir" points to
-	# gitlink file .git in the real worktree.
+	# this is invoked only if the current directory is inside the repository
 	set worktree {}
 	if {[file tail $::_gitdir] eq {.git}} {
+		# the dir containing .git is a worktree if repo allows it
+		# Check that git reports parent as a worktree (gitdir might not allow a worktree)
 		if {[catch {
-			set gitdir_parent [file dirname $::_gitdir]
-			set worktree [git -C $gitdir_parent rev-parse --show-toplevel]
-			set parent_gitdir [git -C $worktree rev-parse --absolute-git-dir]
-			if {$::_gitdir ne $parent_gitdir} {
-				set worktree {}
-			}
-		}]} {
+				set parent [file dirname $::_gitdir]
+				set worktree [git -C $parent rev-parse --show-toplevel]
+			}]} {
 			set worktree {}
 		}
 	} elseif [file exists {gitdir}] {
+		# a worktree gitdir has .gitdir naming worktree/.git
+		# assure git run there reports this dir as the gitdir (links might be broken)
 		if {[catch {
-			set fd_gitdir [open {gitdir} {r}]
-			set gitlink_parent [file dirname [read $fd_gitdir]]
-			catch {close $fd_gitdir}
-			set worktree [git -C $gitlink_parent rev-parse --show-toplevel]
-			set parent_gitdir [git -C $worktree rev-parse --absolute-git-dir]
-			if {$::_gitdir ne $parent_gitdir} {
-				set worktree {}
-			}
-		}]} {
+				set fd_gitdir [open {gitdir} {r}]
+				set worktree [file dirname [read $fd_gitdir]]
+				catch {close $fd_gitdir}
+				set worktree_gitdir [git -C $worktree rev-parse --absolute-git-dir]
+				if {$::_gitdir ne $worktree_gitdir} {
+					set worktree {}
+				}
+			}]} {
 			catch {close $fd_gitdir}
 			set worktree {}
 		}
@@ -1153,7 +1149,7 @@ proc is_gitvars_error {err} {
 	set GIT_DIR {}
 	set GIT_WORK_TREE {}
 	catch {set GIT_DIR $::env(GIT_DIR); set havevars 1}
-	catch {set GIT_WORK_TREE $::env(GIT_WORK_TREE) ; set havevars 1}
+	catch {set GIT_WORK_TREE $::env(GIT_WORK_TREE); set havevars 1}
 
 	if {$havevars} {
 		catch {wm withdraw .}
@@ -1168,9 +1164,7 @@ proc is_gitvars_error {err} {
 
 proc set_gitdir_vars {} {
 	global _gitdir _gitworktree env
-	if {$_gitdir ne {}} {
-		set env(GIT_DIR) $_gitdir
-	}
+	set env(GIT_DIR) $_gitdir
 	if {$_gitworktree ne {}} {
 		set env(GIT_WORK_TREE) $_gitworktree
 	}
@@ -1182,12 +1176,12 @@ proc unset_gitdir_vars {} {
 	catch {unset env(GIT_WORK_TREE)}
 }
 
-# find repository.
+# find repository
 set _gitdir {}
 if {[is_enabled gitdir_discovery]} {
 	if {[catch {
-		set _gitdir [git rev-parse --absolute-git-dir]
-	} err]} {
+			set _gitdir [git rev-parse --absolute-git-dir]
+		} err]} {
 		if {[is_gitvars_error $err]} {
 			exit 1
 		}
@@ -1200,14 +1194,13 @@ if {$_gitdir eq {} && [is_enabled picker]} {
 	unset_gitdir_vars
 	load_config 1
 	apply_config
-	if {![choose_repository::pick]} {
-		exit 1
-	}
+	choose_repository::pick
 	if {[catch {
-		set _gitdir [git rev-parse --absolute-git-dir]
-	} err]} {
+			set _gitdir [git rev-parse --absolute-git-dir]
+		} err]} {
 		catch {wm withdraw .}
-		error_popup [strcat [mc "Unusable repo/worktree:"] " [pwd] "\n\n$err"]
+		error_popup [strcat [mc "Unusable repo/worktree:"] " [pwd] \n\n$err"]
+		exit 1
 	}
 	set picked 1
 }
@@ -1220,9 +1213,9 @@ if {$_gitdir eq {}} {
 
 # find worktree, continue without if not required
 if {[catch {
-	set _gitworktree [git rev-parse --show-toplevel]
-	set _prefix [git rev-parse --show-prefix]
-} err]} {
+		set _gitworktree [git rev-parse --show-toplevel]
+		set _prefix [git rev-parse --show-prefix]
+	} err]} {
 	if {[is_gitvars_error $err]} {
 		exit 1
 	}
@@ -1238,17 +1231,14 @@ if {[is_bare]} {
 }
 
 if {![is_bare]} {
-	if {[catch {
-		cd $_gitworktree
-	} err]} {
+	if {[catch {cd $_gitworktree} err]} {
 		catch {wm withdraw .}
-		error_popup [strcat [mc "Cannot change to discovered worktree: "] \
-			"$_gitworktree" "\n\n$err"]
-		exit 1;
+		error_popup [strcat [mc "No working directory"] " $_gitworktree:\n\n$err"]
+		exit 1
 	}
 } elseif {![is_enabled bare]} {
 	catch {wm withdraw .}
-	error_popup [strcat [mc "Cannot use bare repository:"] "\n\n" $_gitdir]
+	error_popup [strcat [mc "Cannot use bare repository:"] "\n\n$_gitdir"]
 	exit 1
 }
 
@@ -2086,11 +2076,7 @@ proc do_gitk {revs {is_submodule false}} {
 	} else {
 		set pwd [pwd]
 
-		if {!$is_submodule} {
-			if {![is_bare]} {
-				cd $_gitworktree
-			}
-		} else {
+		if {$is_submodule} {
 			cd $current_diff_path
 			if {$revs eq {--}} {
 				set s $file_states($current_diff_path)
@@ -3026,26 +3012,18 @@ proc normalize_relpath {path} {
 	if {$elements ne {}} {
 		return [eval file join $elements]
 	} else {
-		return {./}
+		return {.}
 	}
 }
 
-proc find_path_type {head path} {
-	if {$path eq {./}} {
-		# the root-tree exists in every rev, ls-tree gives data on the contents,
-		# not the type of tree itself. So, if the rev exists, return {tree}
-		if {[catch {set objtype [git ls-tree $head]}]} {
-			set objtype {}
-		} else {
-			set objtype {tree}
-		}
+proc show_parse_err {err} {
+	if {[tk windowingsystem] eq "win32"} {
+		catch {wm withdraw .}
+		error_popup $err
 	} else {
-		# test that the path exists in head, ls-tree gives info on the path only
-		if {[catch {set objtype [git ls-tree {--format=%(objecttype)} $head $path]}]} {
-			set objtype {}
-		}
+		puts stderr $err
 	}
-	return $objtype
+	exit 1
 }
 
 # -- Not a normal commit type invocation?  Do that instead!
@@ -3054,91 +3032,80 @@ switch -- $subcommand {
 browser -
 blame {
 	if {$subcommand eq "blame"} {
-		set subcommand_args {[--line=<num>] <[rev] [--] filename | [--] filename rev>}
-		set required_objtype blob
+		set subcommand_args {[--line=<num>] [rev] [--] <filename>}
+		set required_pathtype blob
 	} else {
-		set subcommand_args {<[rev] [--] directory | [--] directory rev>}
-		set required_objtype tree
+		set subcommand_args {[rev] [--] <dirname>}
+		set required_pathtype tree
 	}
 	set maxargs [llength $subcommand_args]
 	set nargs [llength $argv]
 	if {$nargs < 1 || $nargs > $maxargs} usage
 	set head {}
-	set althead {}
 	set path {}
-	set altpath {}
-	set canswap 1
 	set jump_spec {}
 
-	# assume: [--line=num] [head] [--] path as the possible arguments, in order.
-	# head and path may need a swap later.
-	for {set iarg 0} {$iarg < $nargs} {incr iarg} {
-		set arg [lindex $argv $iarg]
-		if {$arg eq {--}} {
-			# next arg is the path, prevent or FORCE swap?
-			if {$iarg == $nargs - 2} {
-				set canswap 0
-			} elseif {$iarg == $nargs - 3} {
-				set canswap 2
-			} else {
+	set iarg 0
+	foreach a $argv {
+		incr iarg
+		if {$iarg == $nargs} {
+			# final argument is path
+			set path [normalize_relpath [file join $_prefix $a]]
+		} elseif {$a eq {--}} {
+			# allow before required final arg that must be path
+			if {$iarg != $nargs - 1} {
 				usage
 			}
-		} elseif {[regexp {^--line=(\d+)$} $arg arg lnum]} {
+		} elseif {[regexp {^--line=(\d+)$} $a a lnum]} {
 			# --line can only be the first arg
-			if {$iarg != 0 || $maxargs < 4} usage
+			if {$iarg != 1 || $subcommand ne {blame}} usage
 			set jump_spec [list $lnum]
-		} elseif {$iarg == $nargs - 1} {
-			# assume final argument is path
-			set path [normalize_relpath [file join $_prefix $arg]]
-			set althead $arg
 		} elseif {$head eq {}} {
-			# assume the other argument is head
-			set head $arg
-			set altpath [normalize_relpath [file join $_prefix $arg]]
+			set head $a
 		} else {
 			usage
 		}
 	}
 
-	# no swapping allowed if head not given, use current branch (HEAD)
+	# If head not given, use current branch (HEAD),
+	# and blame will use worktree if there is one.
+	set use_worktree 0
 	if {$head eq {}} {
 		load_current_branch
 		set head $current_branch
-		set canswap 0
-	}
-
-	# -- before "rev" arg means we got -- path head
-	if {$canswap == 2} {
-		set head $althead
-		set path $altpath
-		set canswap 0
-	}
-
-	set objtype [find_path_type $head $path]
-	if {$objtype eq {} && $canswap} {
-		set objtype [find_path_type $althead $altpath]
-		if {$objtype ne {}} {
-			set head $althead
-			set path $altpath
+		if {$subcommand eq {blame} && ![is_bare]} {
+			if {![file isfile $path]} {
+				show_parse_err [mc "fatal: no such file '%s' in worktree" $path]
+			}
+			set use_worktree 1
 		}
-	}
-	set current_branch $head
-
-	# check that path exists in head, and objtype matches need
-	if {$objtype ne $required_objtype} {
-		switch -- $required_objtype {
-			tree {set err [strcat \
-				[mc "'%s' is not a directory in rev '%s'" $path $head]]}
-			blob {set err [strcat \
-				[mc "'%s' is not a filename in rev '%s'" $path $head]]}
-		}
-		if {[tk windowingsystem] eq "win32"} {
-			catch {wm withdraw .}
-			error_popup $err
+	} else {
+		if {[catch {
+				set commitid \
+					[git rev-parse --verify --end-of-options \
+					[strcat $head "^{commit}"]]
+			}]} {
+			show_parse_err [mc "fatal: '%s' is not a valid rev'" $head]
 		} else {
-			puts stderr $err
+			set current_branch $head
+		}
+	}
+
+	# check path is known in head, and is file / directory as required
+	set pathtype {}
+	catch {set pathtype [git ls-tree {--format=%(objecttype)} $head $path]}
+	if {$pathtype ne {} && $path eq {.}} {
+		# ls-tree gives contents of root-dir, we need root-dir itself
+		set pathtype {tree}
+	}
+
+	if {$pathtype ne $required_pathtype} {
+		switch -- $required_pathtype {
+			tree {show_parse_err \
+				[mc "'%s' is not a directory in rev '%s'" $path $head]}
+			blob {show_parse_err \
+				[mc "'%s' is not a filename in rev '%s'" $path $head]}
 		}
-		exit 1
 	}
 
 	wm deiconify .
@@ -3146,8 +3113,8 @@ blame {
 	browser {
 		browser::new $head $path
 	}
-	blame {
-		blame::new $head $path $jump_spec
+	blame   {
+		blame::new [expr {$use_worktree ? {} : $head}] $path $jump_spec
 	}
 	}
 	return
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* [PATCH v3 01/12] git-gui: use HEAD as current branch when detached
  2026-05-31 23:02                     ` [PATCH v3 00/12] " Mark Levedahl
@ 2026-05-31 23:02                       ` Mark Levedahl
  2026-05-31 23:02                       ` [PATCH v3 02/12] git-gui: remove unnecessary 'cd $_gitworktree' from do_gitk Mark Levedahl
                                         ` (11 subsequent siblings)
  12 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-31 23:02 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

commit f87a36b697 ("git-gui: use git-branch --show-current", 2024-02-12)
changed git-gui to use git-branch to access refs, rather than directly
reading files as doing the latter is not compatible with the reftable
backend. git branch --show-current reports an empty branch name when the
head is detached, and in this case load_current_branch needs to report
HEAD using special case logic as it did prior to the above commit. Make
it do so.

This addresses an issue with git-gui browser failing with a detached
head.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/git-gui.sh b/git-gui.sh
index 23fe76e498..f70a54a61b 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -670,6 +670,9 @@ proc load_current_branch {} {

 	set current_branch [git branch --show-current]
 	set is_detached [expr [string length $current_branch] == 0]
+	if {$is_detached} {
+		set current_branch {HEAD}
+	}
 }

 auto_load tk_optionMenu
-- 
2.54.0.99.14

^ permalink raw reply related	[flat|nested] 134+ messages in thread

* [PATCH v3 02/12] git-gui: remove unnecessary 'cd $_gitworktree' from do_gitk
  2026-05-31 23:02                     ` [PATCH v3 00/12] " Mark Levedahl
  2026-05-31 23:02                       ` [PATCH v3 01/12] git-gui: use HEAD as current branch when detached Mark Levedahl
@ 2026-05-31 23:02                       ` Mark Levedahl
  2026-05-31 23:02                       ` [PATCH v3 03/12] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE Mark Levedahl
                                         ` (10 subsequent siblings)
  12 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-31 23:02 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

From: Johannes Sixt <j6t@kdbg•org>

In the procedure that invokes Gitk, we have a 'cd $_gitworktree'. Such
a change of the current directory is not necessary, because

- if we have a working tree, then the startup routine has already
  changed the current directory to the root of the working tree, which
  *is* $_gitworktree; or

- if we are in a bare repository, then there is no point in changing
  the current directory anywhere. (And $_gitworktree is empty.)

Signed-off-by: Johannes Sixt <j6t@kdbg•org>
Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index f70a54a61b..52897fbd09 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -2024,11 +2024,7 @@ proc do_gitk {revs {is_submodule false}} {
 
 		set pwd [pwd]
 
-		if {!$is_submodule} {
-			if {![is_bare]} {
-				cd $_gitworktree
-			}
-		} else {
+		if {$is_submodule} {
 			cd $current_diff_path
 			if {$revs eq {--}} {
 				set s $file_states($current_diff_path)
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* [PATCH v3 03/12] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE
  2026-05-31 23:02                     ` [PATCH v3 00/12] " Mark Levedahl
  2026-05-31 23:02                       ` [PATCH v3 01/12] git-gui: use HEAD as current branch when detached Mark Levedahl
  2026-05-31 23:02                       ` [PATCH v3 02/12] git-gui: remove unnecessary 'cd $_gitworktree' from do_gitk Mark Levedahl
@ 2026-05-31 23:02                       ` Mark Levedahl
  2026-05-31 23:02                       ` [PATCH v3 04/12] git-gui: do not change global vars in choose_repository::pick Mark Levedahl
                                         ` (9 subsequent siblings)
  12 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-31 23:02 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

git-gui unconditionally exports _gitdir as GIT_DIR, and _gitworktree as
GIT_WORK_TREE, to the environment, and unconditionally
unsets these environment variables before invoking gitk or git-gui when
a submodule is involved. This export happens even if _gitworktree is
empty, which happens when running from a bare repository. However,
exporting GIT_WORK_TREE as empty is never valid, and causes errors in
git.

GIT_DIR must be exported if the repository is not discoverable from the
worktree (or current directory if there is no worktree). The user might
have configured this.

If there is a worktree, git-gui makes this the current directory.
However, if the repository sets core.worktree, this value can only be
overridden by GIT_WORK_TREE so the latter must be exported.

As we cannot eliminate conditions where either variable is needed, let's
implement a pair of functions to set / unset these variables without
error, and without ever exporting an empty GIT_WORK_TREE.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 37 +++++++++++++++++++++----------------
 1 file changed, 21 insertions(+), 16 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 52897fbd09..2cf14d0dd5 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1125,6 +1125,20 @@ unset argv0dir
 ##
 ## repository setup
 
+proc set_gitdir_vars {} {
+	global _gitdir _gitworktree env
+	set env(GIT_DIR) $_gitdir
+	if {$_gitworktree ne {}} {
+		set env(GIT_WORK_TREE) $_gitworktree
+	}
+}
+
+proc unset_gitdir_vars {} {
+	global env
+	catch {unset env(GIT_DIR)}
+	catch {unset env(GIT_WORK_TREE)}
+}
+
 set picked 0
 if {[catch {
 		set _gitdir $env(GIT_DIR)
@@ -1210,8 +1224,8 @@ if {[lindex $_reponame end] eq {.git}} {
 	set _reponame [lindex $_reponame end]
 }
 
-set env(GIT_DIR) $_gitdir
-set env(GIT_WORK_TREE) $_gitworktree
+# Export the final paths
+set_gitdir_vars
 
 ######################################################################
 ##
@@ -2010,7 +2024,7 @@ proc incr_font_size {font {amt 1}} {
 
 proc do_gitk {revs {is_submodule false}} {
 	global current_diff_path file_states current_diff_side ui_index
-	global _gitdir _gitworktree
+	global _gitworktree
 
 	# -- Always start gitk through whatever we were loaded with.  This
 	#    lets us bypass using shell process on Windows systems.
@@ -2020,8 +2034,6 @@ proc do_gitk {revs {is_submodule false}} {
 	if {$exe eq {}} {
 		error_popup [mc "Couldn't find gitk in PATH"]
 	} else {
-		global env
-
 		set pwd [pwd]
 
 		if {$is_submodule} {
@@ -2049,13 +2061,11 @@ proc do_gitk {revs {is_submodule false}} {
 			# TODO we could make life easier (start up faster?) for gitk
 			# by setting these to the appropriate values to allow gitk
 			# to skip the heuristics to find their proper value
-			unset env(GIT_DIR)
-			unset env(GIT_WORK_TREE)
+			unset_gitdir_vars
 		}
 		safe_exec_bg [concat $cmd $revs "--" "--"]
 
-		set env(GIT_DIR) $_gitdir
-		set env(GIT_WORK_TREE) $_gitworktree
+		set_gitdir_vars
 		cd $pwd
 
 		if {[info exists main_status]} {
@@ -2078,21 +2088,16 @@ proc do_git_gui {} {
 	if {$exe eq {}} {
 		error_popup [mc "Couldn't find git gui in PATH"]
 	} else {
-		global env
-		global _gitdir _gitworktree
-
 		# see note in do_gitk about unsetting these vars when
 		# running tools in a submodule
-		unset env(GIT_DIR)
-		unset env(GIT_WORK_TREE)
+		unset_gitdir_vars
 
 		set pwd [pwd]
 		cd $current_diff_path
 
 		safe_exec_bg [concat $exe gui]
 
-		set env(GIT_DIR) $_gitdir
-		set env(GIT_WORK_TREE) $_gitworktree
+		set_gitdir_vars
 		cd $pwd
 
 		set status_operation [$::main_status \
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* [PATCH v3 04/12] git-gui: do not change global vars in choose_repository::pick
  2026-05-31 23:02                     ` [PATCH v3 00/12] " Mark Levedahl
                                         ` (2 preceding siblings ...)
  2026-05-31 23:02                       ` [PATCH v3 03/12] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE Mark Levedahl
@ 2026-05-31 23:02                       ` Mark Levedahl
  2026-05-31 23:02                       ` [PATCH v3 05/12] git-gui: use --absolute-git-dir Mark Levedahl
                                         ` (8 subsequent siblings)
  12 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-31 23:02 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

The repository picker (choose_repository::pick, AKA pick) on success
always returns with the current directory at the root of the selected
worktree, with the global variable _gitdir holding the name of the
git repository, possibly as a relative path, and _prefix {}. The
worktree root (_gitworktree) is not filled out, and if the selection was
from the "recent" list, no validation has occurred beyond testing that
the worktree root exists. So, repository and worktree validation are
still needed to be sure the new repo + worktree is usable.

pick only supports worktrees with a .git entry in the worktree root, so
git repository and worktree discovery will work starting in the current
directory on return. In cases of error, or user abort, pick exits the
process rather than returning.

So, let's change pick to not alter any global values, with success
indicated by the process returning to the caller. In this case, the
current directory is the worktree root, with a .git entry. The caller
then proceeds with normal discovery to find and validate both repository
and worktree.

With this, pick now returns 1 in the success case, but additional work
would be necessary to return from conditions where 0 should be returned.
Checking this return value would be superfluous.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh                |  7 ++++++-
 lib/choose_repository.tcl | 21 ++++++++-------------
 2 files changed, 14 insertions(+), 14 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 2cf14d0dd5..44914bddcf 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1153,9 +1153,14 @@ if {[catch {
 	load_config 1
 	apply_config
 	choose_repository::pick
-	if {![file isdirectory $_gitdir]} {
+	if {[catch {
+			set _gitdir [git rev-parse --git-dir]
+		} err]} {
+		catch {wm withdraw .}
+		error_popup [strcat [mc "Unusable repo/worktree:"] " [pwd] \n\n$err"]
 		exit 1
 	}
+	set _prefix {}
 	set picked 1
 }
 
diff --git a/lib/choose_repository.tcl b/lib/choose_repository.tcl
index 7e1462a20c..4b06afee93 100644
--- a/lib/choose_repository.tcl
+++ b/lib/choose_repository.tcl
@@ -15,7 +15,7 @@ field w_recentlist ; # Listbox containing recent repositories
 field w_localpath  ; # Entry widget bound to local_path
 
 field done              0 ; # Finished picking the repository?
-field clone_ok      false ; # clone succeeeded
+field pick_ok           0 ; # true if repo pick/clone succeeded
 field local_path       {} ; # Where this repository is locally
 field origin_url       {} ; # Where we are cloning from
 field origin_name  origin ; # What we shall call 'origin'
@@ -220,6 +220,8 @@ constructor pick {} {
 	if {$top eq {.}} {
 		eval destroy [winfo children $top]
 	}
+
+	return $pick_ok
 }
 
 method _center {} {
@@ -327,8 +329,7 @@ method _git_init {} {
 	}
 
 	_append_recentrepos [pwd]
-	set ::_gitdir .git
-	set ::_prefix {}
+	set pick_ok 1
 	return 1
 }
 
@@ -409,6 +410,7 @@ method _do_new2 {} {
 	if {![_git_init $this]} {
 		return
 	}
+	set pick_ok 1
 	set done 1
 }
 
@@ -621,7 +623,7 @@ method _do_clone2 {} {
 	}
 
 	tkwait variable @done
-	if {!$clone_ok} {
+	if {!$pick_ok} {
 		error_popup [mc "Clone failed."]
 		return
 	}
@@ -632,18 +634,12 @@ method _do_clone2_done {ok} {
 	if {$ok} {
 		if {[catch {
 			cd $local_path
-			set ::_gitdir .git
-			set ::_prefix {}
 			_append_recentrepos [pwd]
 		} err]} {
 			set ok 0
 		}
 	}
-	if {!$ok} {
-		set ::_gitdir {}
-		set ::_prefix {}
-	}
-	set clone_ok $ok
+	set pick_ok $ok
 	set done 1
 }
 
@@ -721,8 +717,7 @@ method _do_open2 {} {
 	}
 
 	_append_recentrepos [pwd]
-	set ::_gitdir $actualgit
-	set ::_prefix {}
+	set pick_ok 1
 	set done 1
 }
 
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* [PATCH v3 05/12] git-gui: use --absolute-git-dir
  2026-05-31 23:02                     ` [PATCH v3 00/12] " Mark Levedahl
                                         ` (3 preceding siblings ...)
  2026-05-31 23:02                       ` [PATCH v3 04/12] git-gui: do not change global vars in choose_repository::pick Mark Levedahl
@ 2026-05-31 23:02                       ` Mark Levedahl
  2026-05-31 23:02                       ` [PATCH v3 06/12] git-gui: use rev-parse exclusively to find a repository Mark Levedahl
                                         ` (7 subsequent siblings)
  12 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-31 23:02 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

git-gui uses git rev-parse --git-dir to get the pathname of the
discovered git repository. The returned value can be relative, and is
'.' if the current directory is the top of the repository directory
itself.  git-gui has code to change '.' to [pwd] in this case so that
subsequent logic runs.

But, git rev-parse supports --absolute-git-dir from fac60b8925
("rev-parse: add option for absolute or relative path formatting",
2020-12-13), and included in git 2.31. git-gui requires git >= 2.36, so
this more useful form is always available. Use --absolute-git-dir to
always get an absolute path, avoiding the need for other checks, and
delete the now unneeded code to fix a relative _gitdir.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 14 +++-----------
 1 file changed, 3 insertions(+), 11 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 44914bddcf..8a5f0bd938 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1145,16 +1145,15 @@ if {[catch {
 		set _prefix {}
 		}]
 	&& [catch {
-		# beware that from the .git dir this sets _gitdir to .
-		# and _prefix to the empty string
-		set _gitdir [git rev-parse --git-dir]
+		# beware that from the .git dir this sets _prefix to the empty string
+		set _gitdir [git rev-parse --absolute-git-dir]
 		set _prefix [git rev-parse --show-prefix]
 	} err]} {
 	load_config 1
 	apply_config
 	choose_repository::pick
 	if {[catch {
-			set _gitdir [git rev-parse --git-dir]
+			set _gitdir [git rev-parse --absolute-git-dir]
 		} err]} {
 		catch {wm withdraw .}
 		error_popup [strcat [mc "Unusable repo/worktree:"] " [pwd] \n\n$err"]
@@ -1175,13 +1174,6 @@ if {$hashalgorithm eq "sha1"} {
 	exit 1
 }
 
-# we expand the _gitdir when it's just a single dot (i.e. when we're being
-# run from the .git dir itself) lest the routines to find the worktree
-# get confused
-if {$_gitdir eq "."} {
-	set _gitdir [pwd]
-}
-
 if {![file isdirectory $_gitdir]} {
 	catch {wm withdraw .}
 	error_popup [strcat [mc "Git directory not found:"] "\n\n$_gitdir"]
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* [PATCH v3 06/12] git-gui: use rev-parse exclusively to find a repository
  2026-05-31 23:02                     ` [PATCH v3 00/12] " Mark Levedahl
                                         ` (4 preceding siblings ...)
  2026-05-31 23:02                       ` [PATCH v3 05/12] git-gui: use --absolute-git-dir Mark Levedahl
@ 2026-05-31 23:02                       ` Mark Levedahl
  2026-05-31 23:02                       ` [PATCH v3 07/12] git-gui: use git rev-parse for worktree discovery Mark Levedahl
                                         ` (6 subsequent siblings)
  12 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-31 23:02 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

git-gui attempts to use env(GIT_DIR) directly as the git repository,
accepting GIT_DIR if it is a directory. Only if that fails is git
rev-parse used to discover the repository.  But, this avoids all of
git-core's validity checking on a repository, thus possibly deferring an
error to a later step, possibly unexpected. Repository validation should
be part of initial setup so that later processing does not need error
trapping for configuration errors.

Let's just invoke rev-parse so all error checking is done.

While here, let's cleanup the error handling.

Stop if an error occurs and the user set GIT_DIR or GIT_WORK_TREE.
Use of either or both of those variables is supported by git, but their
use also means the user has taken responsibility that they are correct,
so a failure is something the user must address.

Otherwise on error, continue the existing behavior and show the
repository picker. But, let's move the possible invocation of
repository_chooser::pick to a separate code block. This permits adding
separate conditions on using pick independent of repository discovery, and
will be exploited later in the series.  Note that the picker always
returns with the current directory in the root of a worktree with the
git repository is in the .git subdirectory.  The variable "picked" is
used by git-gui to automatically execute the "Explore Working Copy" menu
item after the repository picker is run.  This is controlled by config
variable gui.autoexplore, and happens after all discovery is complete.

Remove a later check on whether _gitdir is a directory: that code
cannot be reached without rev-parse already validating the repository.

_prefix is set as part of worktree discovery, but must be {} if not
running with a worktree. Initialze this as {} along with other global
variables, this is the correct value is no worktree is found.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 49 ++++++++++++++++++++++++++++++++++---------------
 1 file changed, 34 insertions(+), 15 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 8a5f0bd938..aed3bc7200 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -374,6 +374,7 @@ set _gitdir {}
 set _gitworktree {}
 set _isbare {}
 set _githtmldir {}
+set _prefix {}
 set _reponame {}
 set _shellpath {@@SHELL_PATH@@}

@@ -1125,6 +1126,24 @@ unset argv0dir
 ##
 ## repository setup

+proc is_gitvars_error {err} {
+	set havevars 0
+	set GIT_DIR {}
+	set GIT_WORK_TREE {}
+	catch {set GIT_DIR $::env(GIT_DIR); set havevars 1}
+	catch {set GIT_WORK_TREE $::env(GIT_WORK_TREE); set havevars 1}
+
+	if {$havevars} {
+		catch {wm withdraw .}
+		error_popup [strcat [mc "Invalid configuration:"] \
+		   "\n" "GIT_DIR: " $GIT_DIR \
+		   "\n" "GIT_WORK_TREE: " $GIT_WORK_TREE \
+			"\n\n$err"]
+		return 1
+	}
+	return 0
+}
+
 proc set_gitdir_vars {} {
 	global _gitdir _gitworktree env
 	set env(GIT_DIR) $_gitdir
@@ -1139,16 +1158,22 @@ proc unset_gitdir_vars {} {
 	catch {unset env(GIT_WORK_TREE)}
 }

+# find repository
+set _gitdir {}
+if {$_gitdir eq {}} {
+	if {[catch {
+			set _gitdir [git rev-parse --absolute-git-dir]
+		} err]} {
+		if {[is_gitvars_error $err]} {
+			exit 1
+		}
+		set _gitdir {}
+	}
+}
+
 set picked 0
-if {[catch {
-		set _gitdir $env(GIT_DIR)
-		set _prefix {}
-		}]
-	&& [catch {
-		# beware that from the .git dir this sets _prefix to the empty string
-		set _gitdir [git rev-parse --absolute-git-dir]
-		set _prefix [git rev-parse --show-prefix]
-	} err]} {
+if {$_gitdir eq {}} {
+	unset_gitdir_vars
 	load_config 1
 	apply_config
 	choose_repository::pick
@@ -1159,7 +1184,6 @@ if {[catch {
 		error_popup [strcat [mc "Unusable repo/worktree:"] " [pwd] \n\n$err"]
 		exit 1
 	}
-	set _prefix {}
 	set picked 1
 }

@@ -1174,11 +1198,6 @@ if {$hashalgorithm eq "sha1"} {
 	exit 1
 }

-if {![file isdirectory $_gitdir]} {
-	catch {wm withdraw .}
-	error_popup [strcat [mc "Git directory not found:"] "\n\n$_gitdir"]
-	exit 1
-}
 # _gitdir exists, so try loading the config
 load_config 0
 apply_config
-- 
2.54.0.99.14

^ permalink raw reply related	[flat|nested] 134+ messages in thread

* [PATCH v3 07/12] git-gui: use git rev-parse for worktree discovery
  2026-05-31 23:02                     ` [PATCH v3 00/12] " Mark Levedahl
                                         ` (5 preceding siblings ...)
  2026-05-31 23:02                       ` [PATCH v3 06/12] git-gui: use rev-parse exclusively to find a repository Mark Levedahl
@ 2026-05-31 23:02                       ` Mark Levedahl
  2026-05-31 23:02                       ` [PATCH v3 08/12] git-gui: simplify [is_bare] to report if a worktree is known Mark Levedahl
                                         ` (5 subsequent siblings)
  12 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-31 23:02 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

git gui uses a combination of tcl code and git invocations to determine
the worktree and the location with respect to the worktree root
(_prefix). But, git rev-parse provides all of this information directly,
and assures full error and configuration checking are done by git
itself. The entirety of discovery in normal configurations involves

	git rev-parse --show-toplevel (gets worktree root)
	git rev-parse --show-prefix (shows location wrt the root)

An error thrown on either of these lines means the worktree discovered
by git is unusable, or git did not discover a worktree because the
current directory is inside the repository. If the user has defined
GIT_DIR or GIT_WORK_TREE, this is a user configuration error and git-gui
should stop.

Otherwise, the blame or browser subcommands can be used without a
worktree.

A separate error might occur when changing to the root of the discovered
worktree. The cause would be file system related and completely outside
of git's control, so trap that independently.

Discovery of the repository and the worktree must be guarded to trap
errors: the intent is that any configuration problems are caught during
discovery, and later processing need not include error trapping and
recovery. So, move all worktree discovery code to be immediately after
repository discovery.

This does move configuration loading to occur after worktree discovery
rather than before. None of the code executed in worktree discovery has
any option controlled by a git-gui configuration variable, so no impact
is expected. git itself will always read the repository configuration,
including worktree specific configuration data if that exists, so this
is unaffected by when git-gui loads its own config data. Also, we cannot
be sure the worktree dependent configuration can be loaded before
full discovery is complete.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 61 ++++++++++++++++++++++++------------------------------
 1 file changed, 27 insertions(+), 34 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index aed3bc7200..53b6859345 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1187,6 +1187,33 @@ if {$_gitdir eq {}} {
 	set picked 1
 }

+# find worktree, continue without if not required
+if {[catch {
+		set _gitworktree [git rev-parse --show-toplevel]
+		set _prefix [git rev-parse --show-prefix]
+	} err]} {
+	if {[is_gitvars_error $err]} {
+		exit 1
+	}
+	set _gitworktree {}
+	set _prefix {}
+}
+
+if {![is_bare]} {
+	if {[catch {cd $_gitworktree} err]} {
+		catch {wm withdraw .}
+		error_popup [strcat [mc "No working directory"] " $_gitworktree:\n\n$err"]
+		exit 1
+	}
+} elseif {![is_enabled bare]} {
+	catch {wm withdraw .}
+	error_popup [strcat [mc "Cannot use bare repository:"] "\n\n$_gitdir"]
+	exit 1
+}
+
+# repository and worktree config are complete, export them
+set_gitdir_vars
+
 # Use object format as hash algorithm (either "sha1" or "sha256")
 set hashalgorithm [git rev-parse --show-object-format]
 if {$hashalgorithm eq "sha1"} {
@@ -1202,37 +1229,6 @@ if {$hashalgorithm eq "sha1"} {
 load_config 0
 apply_config

-set _gitworktree [git rev-parse --show-toplevel]
-
-if {$_prefix ne {}} {
-	if {$_gitworktree eq {}} {
-		regsub -all {[^/]+/} $_prefix ../ cdup
-	} else {
-		set cdup $_gitworktree
-	}
-	if {[catch {cd $cdup} err]} {
-		catch {wm withdraw .}
-		error_popup [strcat [mc "Cannot move to top of working directory:"] "\n\n$err"]
-		exit 1
-	}
-	set _gitworktree [pwd]
-	unset cdup
-} elseif {![is_enabled bare]} {
-	if {[is_bare]} {
-		catch {wm withdraw .}
-		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n$_gitdir"]
-		exit 1
-	}
-	if {$_gitworktree eq {}} {
-		set _gitworktree [file dirname $_gitdir]
-	}
-	if {[catch {cd $_gitworktree} err]} {
-		catch {wm withdraw .}
-		error_popup [strcat [mc "No working directory"] " $_gitworktree:\n\n$err"]
-		exit 1
-	}
-	set _gitworktree [pwd]
-}
 set _reponame [file split [file normalize $_gitdir]]
 if {[lindex $_reponame end] eq {.git}} {
 	set _reponame [lindex $_reponame end-1]
@@ -1240,9 +1236,6 @@ if {[lindex $_reponame end] eq {.git}} {
 	set _reponame [lindex $_reponame end]
 }

-# Export the final paths
-set_gitdir_vars
-
 ######################################################################
 ##
 ## global init
-- 
2.54.0.99.14

^ permalink raw reply related	[flat|nested] 134+ messages in thread

* [PATCH v3 08/12] git-gui: simplify [is_bare] to report if a worktree is known
  2026-05-31 23:02                     ` [PATCH v3 00/12] " Mark Levedahl
                                         ` (6 preceding siblings ...)
  2026-05-31 23:02                       ` [PATCH v3 07/12] git-gui: use git rev-parse for worktree discovery Mark Levedahl
@ 2026-05-31 23:02                       ` Mark Levedahl
  2026-05-31 23:02                       ` [PATCH v3 09/12] git-gui: try harder to find worktree from gitdir Mark Levedahl
                                         ` (4 subsequent siblings)
  12 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-31 23:02 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

git-gui includes proc is_bare, used in several places to make decisions
on whether a worktree exists, but also in discovery to tell if a
worktree can be supported.

But, is_bare is out of date with regard to multiple worktrees, safe
repository guards, and possibly other relevant features known to git
rev-parse. Also, is_bare caches its result on the first call, so is not
useful if a later step in the discovery process finds a worktree.

So, simplify is_bare to report whether git-gui has a worktree or is
working only from a repository.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 25 +------------------------
 1 file changed, 1 insertion(+), 24 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 53b6859345..abee222903 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -372,7 +372,6 @@ if {[tk windowingsystem] eq "aqua"} {
 set _appname {Git Gui}
 set _gitdir {}
 set _gitworktree {}
-set _isbare {}
 set _githtmldir {}
 set _prefix {}
 set _reponame {}
@@ -524,29 +523,7 @@ proc get_config {name} {
 }
 
 proc is_bare {} {
-	global _isbare
-	global _gitdir
-	global _gitworktree
-
-	if {$_isbare eq {}} {
-		if {[catch {
-			set _bare [git rev-parse --is-bare-repository]
-			switch  -- $_bare {
-			true { set _isbare 1 }
-			false { set _isbare 0}
-			default { throw }
-			}
-		}]} {
-			if {[is_config_true core.bare]
-				|| ($_gitworktree eq {}
-					&& [lindex [file split $_gitdir] end] ne {.git})} {
-				set _isbare 1
-			} else {
-				set _isbare 0
-			}
-		}
-	}
-	return $_isbare
+	return [expr {$::_gitworktree eq {}}]
 }
 
 ######################################################################
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* [PATCH v3 09/12] git-gui: try harder to find worktree from gitdir
  2026-05-31 23:02                     ` [PATCH v3 00/12] " Mark Levedahl
                                         ` (7 preceding siblings ...)
  2026-05-31 23:02                       ` [PATCH v3 08/12] git-gui: simplify [is_bare] to report if a worktree is known Mark Levedahl
@ 2026-05-31 23:02                       ` Mark Levedahl
  2026-05-31 23:02                       ` [PATCH v3 10/12] git-gui: allow specifying path '.' to the browser Mark Levedahl
                                         ` (3 subsequent siblings)
  12 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-31 23:02 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

git-gui, since 87cd09f43e ("git-gui: work from the .git dir",
2010-01-23), has had the intent to allow starting from inside a
repository, then switching to the parent directory if that is a valid
worktree.

This certainly hasn't worked since 2d92ab32fd ("rev-parse: make
--show-toplevel without a worktree an error", 2019-11-19) in git, but
breaking this git-gui feature was unintentional.

There are (at least) 3 cases where the gitdir can tell us where the
worktree is, and we would like all to work:

- core.worktree is set, and points to a valid worktree. This is already
  handled  by git rev-parse --show-toplevel, even when not in the worktree.
  There is nothing more to do in this case.

- the gitdir is embedded in a worktree as subdirectory .git. The parent
  is (or at least should be) a valid worktree. This worked long ago.

- the gitdir is a worktree specific directory (under
  <mainrepo>/worktrees/worktree_name), within which there is a file
  "gitdir" pointing to .git in the worktree. git gui never learned to
  handle this case.

Let's handle the latter two cases. Always check that the discovered
worktree is valid and points to the already discovered gitdir according
to git rev-parse. This avoids issues that may arise because we are
discovering from the gitdir up, rather than the worktree down, and file
system non-posix behavior or misconfiguration of git might cause
confusion.  For instance, a manually moved worktree might not be where
the gitdir points, or the gitdir might be configured with
core.bare=true.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 38 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 38 insertions(+)

diff --git a/git-gui.sh b/git-gui.sh
index abee222903..f48d929da8 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1103,6 +1103,37 @@ unset argv0dir
 ##
 ## repository setup
 
+proc find_worktree_from_gitdir {} {
+	# this is invoked only if the current directory is inside the repository
+	set worktree {}
+	if {[file tail $::_gitdir] eq {.git}} {
+		# the dir containing .git is a worktree if repo allows it
+		# Check that git reports parent as a worktree (gitdir might not allow a worktree)
+		if {[catch {
+				set parent [file dirname $::_gitdir]
+				set worktree [git -C $parent rev-parse --show-toplevel]
+			}]} {
+			set worktree {}
+		}
+	} elseif [file exists {gitdir}] {
+		# a worktree gitdir has .gitdir naming worktree/.git
+		# assure git run there reports this dir as the gitdir (links might be broken)
+		if {[catch {
+				set fd_gitdir [open {gitdir} {r}]
+				set worktree [file dirname [read $fd_gitdir]]
+				catch {close $fd_gitdir}
+				set worktree_gitdir [git -C $worktree rev-parse --absolute-git-dir]
+				if {$::_gitdir ne $worktree_gitdir} {
+					set worktree {}
+				}
+			}]} {
+			catch {close $fd_gitdir}
+			set worktree {}
+		}
+	}
+	return $worktree
+}
+
 proc is_gitvars_error {err} {
 	set havevars 0
 	set GIT_DIR {}
@@ -1176,6 +1207,13 @@ if {[catch {
 	set _prefix {}
 }
 
+if {[is_bare]} {
+	# Maybe we are in an embedded or worktree specific gitdir
+	if {[set _gitworktree [find_worktree_from_gitdir]] ne {}} {
+		set _prefix {}
+	}
+}
+
 if {![is_bare]} {
 	if {[catch {cd $_gitworktree} err]} {
 		catch {wm withdraw .}
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* [PATCH v3 10/12] git-gui: allow specifying path '.' to the browser
  2026-05-31 23:02                     ` [PATCH v3 00/12] " Mark Levedahl
                                         ` (8 preceding siblings ...)
  2026-05-31 23:02                       ` [PATCH v3 09/12] git-gui: try harder to find worktree from gitdir Mark Levedahl
@ 2026-05-31 23:02                       ` Mark Levedahl
  2026-05-31 23:02                       ` [PATCH v3 11/12] git-gui: check browser/blame arguments carefully Mark Levedahl
                                         ` (2 subsequent siblings)
  12 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-31 23:02 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

Invoking "git-gui browser rev ." should show the file browser for the
commitish rev, starting at the current directory. When the current
directory is the working tree root, this errors out in normalize_relpath
because the '.' is removed, yielding an empty list as argument to [file
join ...]. git ls-tree (underlying the browser) accepts '.', so use that
as the value when in the root.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/git-gui.sh b/git-gui.sh
index f48d929da8..16d6b3051a 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -2993,7 +2993,11 @@ proc normalize_relpath {path} {
 		}
 		lappend elements $item
 	}
-	return [eval file join $elements]
+	if {$elements ne {}} {
+		return [eval file join $elements]
+	} else {
+		return {.}
+	}
 }
 
 # -- Not a normal commit type invocation?  Do that instead!
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* [PATCH v3 11/12] git-gui: check browser/blame arguments carefully
  2026-05-31 23:02                     ` [PATCH v3 00/12] " Mark Levedahl
                                         ` (9 preceding siblings ...)
  2026-05-31 23:02                       ` [PATCH v3 10/12] git-gui: allow specifying path '.' to the browser Mark Levedahl
@ 2026-05-31 23:02                       ` Mark Levedahl
  2026-05-31 23:02                       ` [PATCH v3 12/12] git-gui: add gui and pick as explicit subcommands Mark Levedahl
  2026-06-02 17:34                       ` [PATCH v3 00/12] Improve git gui operation without a worktree Johannes Sixt
  12 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-31 23:02 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

git gui offers two related commands, browser and blame, that provide
graphical interfaces driven by git ls-tree and git blame. As such, the
arguments to git-gui need to satisfy those two git commands. But,
git-gui does not assure this leading to confusing or incorrect results.
For instance 'git browser <non-existent path>' shows a blank browser
window rather than error message.

Also, commit 3e45ee1ef2 ("git-gui: Smarter command line parsing for
browser, blame", 2007-05-08) implemented code to allow giving path
before rev on the command line, and unconditionally uses the worktree to
disambiguate. As a result, the following command run in a current
git-gui checkout of the master branch shows the master branch version of
blame.tcl, when none should be shown as that file does not exist in
gitgui-0.6.0.

  git gui blame lib/blame.tcl gitgui-0.6.0

This 'file before rev' feature in git-gui mirrors ideas considered when
git's user interface was very young, but no such feature is documented
for any git command.  Rather than try to fix an idea git itself
rejected, let's just remove this broken and hopefully unused feature.

git-gui browser|blame both accept 'rev' and 'path' as command line
arguments.  rev defaults to 'HEAD' if not given, while path must be
given. path names a directory tree to ls-tree or a file to blame. path
must exist in rev for ls-tree and for blame.  In addition git blame will
include uncommitted changes from the worktree file at 'path' if rev is
not given (thus defaulting to HEAD), but still requires that the file
exists in HEAD.

So, let's clean up the parser to check that the arguments are usable.
- give a full synopsis, including '--' that may be used to separate rev and
  path. (as path is the required final arg, -- gives no extra info)
- explicitly check the number of arguments
- use rev-parse to assure a user supplied rev is valid
- use ls-tree to assure that path exists in rev
- for blame only, with no rev given and a worktree existing, also assure
  that path points to a file in the worktree

With these changes, error messages are thrown by the parser if the path
or rev are not known: no blank or erroneous displays are created. Also,
this avoids accessing the worktree except in the specific use case
supported by blame / git-blame, meaning browser|blame now also work
without a worktree.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 122 +++++++++++++++++++++++++++--------------------------
 1 file changed, 63 insertions(+), 59 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 16d6b3051a..22939215a6 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -3000,101 +3000,105 @@ proc normalize_relpath {path} {
 	}
 }
 
+proc show_parse_err {err} {
+	if {[tk windowingsystem] eq "win32"} {
+		catch {wm withdraw .}
+		error_popup $err
+	} else {
+		puts stderr $err
+	}
+	exit 1
+}
+
 # -- Not a normal commit type invocation?  Do that instead!
 #
 switch -- $subcommand {
 browser -
 blame {
 	if {$subcommand eq "blame"} {
-		set subcommand_args {[--line=<num>] rev? path}
+		set subcommand_args {[--line=<num>] [rev] [--] <filename>}
+		set required_pathtype blob
 	} else {
-		set subcommand_args {rev? path}
+		set subcommand_args {[rev] [--] <dirname>}
+		set required_pathtype tree
 	}
-	if {$argv eq {}} usage
+	set maxargs [llength $subcommand_args]
+	set nargs [llength $argv]
+	if {$nargs < 1 || $nargs > $maxargs} usage
 	set head {}
 	set path {}
 	set jump_spec {}
-	set is_path 0
-	foreach a $argv {
-		set p [file join $_prefix $a]
 
-		if {$is_path || [file exists $p]} {
-			if {$path ne {}} usage
-			set path [normalize_relpath $p]
-			break
+	set iarg 0
+	foreach a $argv {
+		incr iarg
+		if {$iarg == $nargs} {
+			# final argument is path
+			set path [normalize_relpath [file join $_prefix $a]]
 		} elseif {$a eq {--}} {
-			if {$path ne {}} {
-				if {$head ne {}} usage
-				set head $path
-				set path {}
+			# allow before required final arg that must be path
+			if {$iarg != $nargs - 1} {
+				usage
 			}
-			set is_path 1
 		} elseif {[regexp {^--line=(\d+)$} $a a lnum]} {
-			if {$jump_spec ne {} || $head ne {}} usage
+			# --line can only be the first arg
+			if {$iarg != 1 || $subcommand ne {blame}} usage
 			set jump_spec [list $lnum]
 		} elseif {$head eq {}} {
-			if {$head ne {}} usage
 			set head $a
-			set is_path 1
 		} else {
 			usage
 		}
 	}
-	unset is_path
 
-	if {$head ne {} && $path eq {}} {
-		if {[string index $head 0] eq {/}} {
-			set path [normalize_relpath $head]
-			set head {}
+	# If head not given, use current branch (HEAD),
+	# and blame will use worktree if there is one.
+	set use_worktree 0
+	if {$head eq {}} {
+		load_current_branch
+		set head $current_branch
+		if {$subcommand eq {blame} && ![is_bare]} {
+			if {![file isfile $path]} {
+				show_parse_err [mc "fatal: no such file '%s' in worktree" $path]
+			}
+			set use_worktree 1
+		}
+	} else {
+		if {[catch {
+				set commitid \
+					[git rev-parse --verify --end-of-options \
+					[strcat $head "^{commit}"]]
+			}]} {
+			show_parse_err [mc "fatal: '%s' is not a valid rev'" $head]
 		} else {
-			set path [normalize_relpath $_prefix$head]
-			set head {}
+			set current_branch $head
 		}
 	}
 
-	if {$head eq {}} {
-		load_current_branch
-	} else {
-		if {[regexp [string map "@@ [expr $hashlength - 1]" {^[0-9a-f]{1,@@}$}] $head]} {
-			if {[catch {
-					set head [git rev-parse --verify $head]
-				} err]} {
-				if {[tk windowingsystem] eq "win32"} {
-					tk_messageBox -icon error -title [mc Error] -message $err
-				} else {
-					puts stderr $err
-				}
-				exit 1
-			}
+	# check path is known in head, and is file / directory as required
+	set pathtype {}
+	catch {set pathtype [git ls-tree {--format=%(objecttype)} $head $path]}
+	if {$pathtype ne {} && $path eq {.}} {
+		# ls-tree gives contents of root-dir, we need root-dir itself
+		set pathtype {tree}
+	}
+
+	if {$pathtype ne $required_pathtype} {
+		switch -- $required_pathtype {
+			tree {show_parse_err \
+				[mc "'%s' is not a directory in rev '%s'" $path $head]}
+			blob {show_parse_err \
+				[mc "'%s' is not a filename in rev '%s'" $path $head]}
 		}
-		set current_branch $head
 	}
 
 	wm deiconify .
 	switch -- $subcommand {
 	browser {
-		if {$jump_spec ne {}} usage
-		if {$head eq {}} {
-			if {$path ne {} && [file isdirectory $path]} {
-				set head $current_branch
-			} else {
-				set head $path
-				set path {}
-			}
-		}
 		browser::new $head $path
 	}
 	blame   {
-		if {$head eq {} && ![file exists $path]} {
-			catch {wm withdraw .}
-			tk_messageBox \
-				-icon error \
-				-type ok \
-				-title [mc "git-gui: fatal error"] \
-				-message [mc "fatal: cannot stat path %s: No such file or directory" $path]
-			exit 1
-		}
-		blame::new $head $path $jump_spec
+		blame::new [expr {$use_worktree ? {} : $head}] $path $jump_spec
 	}
 	}
 	return
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* [PATCH v3 12/12] git-gui: add gui and pick as explicit subcommands
  2026-05-31 23:02                     ` [PATCH v3 00/12] " Mark Levedahl
                                         ` (10 preceding siblings ...)
  2026-05-31 23:02                       ` [PATCH v3 11/12] git-gui: check browser/blame arguments carefully Mark Levedahl
@ 2026-05-31 23:02                       ` Mark Levedahl
  2026-06-02 17:34                       ` [PATCH v3 00/12] Improve git gui operation without a worktree Johannes Sixt
  12 siblings, 0 replies; 134+ messages in thread
From: Mark Levedahl @ 2026-05-31 23:02 UTC (permalink / raw)
  To: git; +Cc: j6t, egg_mushroomcow, bootaina702, Mark Levedahl

git-gui accepts subcommands blame | browser | citool, and assumes the
subcommand is 'gui' if none is actually given, But, git-gui also has a
repository picker (choose_repository::pick) that can create a new
repository + worktree, or choose an existing one, switch to that, and
the run the gui. The user has no direct control over invoking the
picker, instead the picker is triggered by failure in the repository /
worktree discovery process: this includes being started in a directory
not controlled by git, which is probably the intended use case.

The picker can appear when the user has no intention of creating a new
worktree, and the user cannot use the picker to create a new worktree
inside another.

So, add two explicit subcommands:
    gui  - Run the gui if repository/worktree discovery succeeds, or die
           with an error message, but never run the picker.
    pick - First run the picker, regardless, then start the gui in
           the chosen worktree.

Nothing in this changes the prior behavior, the alternates above must be
explicitly selected to see any change.

Signed-off-by: Mark Levedahl <mlevedahl@gmail•com>
---
 git-gui.sh | 25 +++++++++++++++++++++----
 1 file changed, 21 insertions(+), 4 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 22939215a6..933e72c9b2 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1024,6 +1024,8 @@ proc load_config {include_global} {
 ##
 ## feature option selection
 
+enable_option picker
+enable_option gitdir_discovery
 if {[regexp {^git-(.+)$} [file tail $argv0] _junk subcommand]} {
 	unset _junk
 } else {
@@ -1035,6 +1037,9 @@ if {$subcommand eq {gui.sh}} {
 if {$subcommand eq {gui} && [llength $argv] > 0} {
 	set subcommand [lindex $argv 0]
 	set argv [lrange $argv 1 end]
+	if {$subcommand eq {gui}} {
+		disable_option picker
+	}
 }
 
 enable_option multicommit
@@ -1050,6 +1055,7 @@ blame {
 	disable_option multicommit
 	disable_option branch
 	disable_option transport
+	disable_option picker
 }
 citool {
 	enable_option singlecommit
@@ -1058,6 +1064,7 @@ citool {
 	disable_option multicommit
 	disable_option branch
 	disable_option transport
+	disable_option picker
 
 	while {[llength $argv] > 0} {
 		set a [lindex $argv 0]
@@ -1080,6 +1087,9 @@ citool {
 		set argv [lrange $argv 1 end]
 	}
 }
+pick {
+	disable_option gitdir_discovery
+}
 }
 
 ######################################################################
@@ -1168,7 +1178,7 @@ proc unset_gitdir_vars {} {
 
 # find repository
 set _gitdir {}
-if {$_gitdir eq {}} {
+if {[is_enabled gitdir_discovery]} {
 	if {[catch {
 			set _gitdir [git rev-parse --absolute-git-dir]
 		} err]} {
@@ -1180,7 +1190,7 @@ if {$_gitdir eq {}} {
 }
 
 set picked 0
-if {$_gitdir eq {}} {
+if {$_gitdir eq {} && [is_enabled picker]} {
 	unset_gitdir_vars
 	load_config 1
 	apply_config
@@ -1195,6 +1205,12 @@ if {$_gitdir eq {}} {
 	set picked 1
 }
 
+if {$_gitdir eq {}} {
+	catch {wm withdraw .}
+	error_popup [strcat [mc "Git directory not found:"] "\n\n$err"]
+	exit 1
+}
+
 # find worktree, continue without if not required
 if {[catch {
 		set _gitworktree [git rev-parse --show-toplevel]
@@ -3104,14 +3120,15 @@ blame {
 	return
 }
 citool -
-gui {
+gui -
+pick {
 	if {[llength $argv] != 0} {
 		usage
 	}
 	# fall through to setup UI for commits
 }
 default {
-	set err "[mc usage:] $argv0 \[{blame|browser|citool}\]"
+	set err "[mc usage:] $argv0 \[{blame|browser|citool|gui|pick}\]"
 	if {[tk windowingsystem] eq "win32"} {
 		wm withdraw .
 		tk_messageBox -icon error -message $err \
-- 
2.54.0.99.14


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v3 00/12] Improve git gui operation without a worktree
  2026-05-31 23:02                     ` [PATCH v3 00/12] " Mark Levedahl
                                         ` (11 preceding siblings ...)
  2026-05-31 23:02                       ` [PATCH v3 12/12] git-gui: add gui and pick as explicit subcommands Mark Levedahl
@ 2026-06-02 17:34                       ` Johannes Sixt
  2026-06-02 18:54                         ` Mark Levedahl
  12 siblings, 1 reply; 134+ messages in thread
From: Johannes Sixt @ 2026-06-02 17:34 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 01.06.26 um 01:02 schrieb Mark Levedahl:
> git gui has a number of inter-related problems that result in problems
> during startup from anything but a checked out worktree pointing at a
> valid git repository. Some of the symptoms are:
> - blame / browser subcommands, and launching gitk, are intended to be
>   useful without a worktree, but fail to work.
> - unlike git, git-gui is supposed to use the parent directory as a
>   worktree if started from the .git subdirectory in the very common
>   single worktree + embedded git repository format. This does not
>   work.
> - git-gui includes a repository picker allowing a user to select a
>   worktree from a list and/or start a new repo+worktree: this dialog can
>   appear at unexpected times, masking useful error feedback on
>   configuration problems.
> 
> This patch series addresses the above issues, substantially rewriting
> the initial repository/worktree process to rely upon git rev-parse so
> that git's knowledge of access rules, repository configuration, and use
> of GIT_DIR / GIT_WORK_TREE (or git --gitdir / --work-tree) is used
> throughout, replacing code largely based upon what git did in 2008. This
> also means that git gui will naturally gain any new rules implmented in
> git-core.
> 
> With this, git-gui only exports GIT_WORK_TREE when non-empty.
> GIT_WORK_TREE is needed, and must be exported, if the user is overriding
> core.worktree in the git repository. But, GIT_WORK_TREE cannot be used
> to specify the lack of a worktree, so exporting an empty GIT_WORK_TREE
> is one of the problems fixed by this series.
> 
> v3 of this series addresses j6t's review of v2, with some reordering of
> patches (1 from j6t added, patch #8 moved to #1), adds another rewrite
> of the browser / blame parser that eliminates the notion of path before
> rev on the command line, blame works correctly with a user modified file
> in the worktree. Clarification is added on the need for GTI_WORK_TREE,
> and the logic in finding a worktree from the gitdir is simplified.
> 
> Johannes Sixt (1):
>   git-gui: remove unnecessary 'cd $_gitworktree' from do_gitk
> 
> Mark Levedahl (11):
>   git-gui: use HEAD as current branch when detached
>   git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE
>   git-gui: do not change global vars in choose_repository::pick
>   git-gui: use --absolute-git-dir
>   git-gui: use rev-parse exclusively to find a repository
>   git-gui: use git rev-parse for worktree discovery
>   git-gui: simplify [is_bare] to report if a worktree is known
>   git-gui: try harder to find worktree from gitdir
>   git-gui: allow specifying path '.' to the browser
>   git-gui: check browser/blame arguments carefully
>   git-gui: add gui and pick as explicit subcommands
> 
>  git-gui.sh                | 377 ++++++++++++++++++++++----------------
>  lib/choose_repository.tcl |  21 +--
>  2 files changed, 223 insertions(+), 175 deletions(-)

This round looks excellent! Thank you very much!

While queuing, I applied the small fixup below to 03/12.

-- Hannes

diff --git a/git-gui.sh b/git-gui.sh
index 933e72c9b255..15dd2b3a84cc 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -2064,7 +2064,6 @@ proc incr_font_size {font {amt 1}} {
 
 proc do_gitk {revs {is_submodule false}} {
 	global current_diff_path file_states current_diff_side ui_index
-	global _gitworktree
 
 	# -- Always start gitk through whatever we were loaded with.  This
 	#    lets us bypass using shell process on Windows systems.


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v3 00/12] Improve git gui operation without a worktree
  2026-06-02 17:34                       ` [PATCH v3 00/12] Improve git gui operation without a worktree Johannes Sixt
@ 2026-06-02 18:54                         ` Mark Levedahl
  2026-06-02 21:05                           ` Johannes Sixt
  0 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-06-02 18:54 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: egg_mushroomcow, bootaina702, git



On 6/2/26 1:34 PM, Johannes Sixt wrote:
>>  2 files changed, 223 insertions(+), 175 deletions(-)
> This round looks excellent! Thank you very much!
>
> While queuing, I applied the small fixup below to 03/12.
>
> -- Hannes
>
> diff --git a/git-gui.sh b/git-gui.sh
> index 933e72c9b255..15dd2b3a84cc 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -2064,7 +2064,6 @@ proc incr_font_size {font {amt 1}} {
>  
>  proc do_gitk {revs {is_submodule false}} {
>  	global current_diff_path file_states current_diff_side ui_index
> -	global _gitworktree
>  
>  	# -- Always start gitk through whatever we were loaded with.  This
>  	#    lets us bypass using shell process on Windows systems.
>

Oops, missed that. Thanks for catching.

You have duplicate signoffs on 02/12 in your repo.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v3 00/12] Improve git gui operation without a worktree
  2026-06-02 18:54                         ` Mark Levedahl
@ 2026-06-02 21:05                           ` Johannes Sixt
  0 siblings, 0 replies; 134+ messages in thread
From: Johannes Sixt @ 2026-06-02 21:05 UTC (permalink / raw)
  To: Mark Levedahl; +Cc: egg_mushroomcow, bootaina702, git

Am 02.06.26 um 20:54 schrieb Mark Levedahl:
> You have duplicate signoffs on 02/12 in your repo.
That's intentional. The patch went from my hands to yours, then back to
mine.

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

[parent not found: <20260509133756.1367-1-egg_mushroomcow@foxmail.com>]

* [PATCH v7 1/3] git-gui: restructure repository startup
       [not found]             ` <20260509133756.1367-1-egg_mushroomcow@foxmail.com>
@ 2026-05-09 13:37               ` Shroom Moo
  2026-05-15  8:26                 ` Johannes Sixt
  2026-05-09 13:37               ` [PATCH v7 2/3] git-gui: disable gitk visualization when no worktree available Shroom Moo
  2026-05-09 13:37               ` [PATCH v7 3/3] git-gui: handle GIT_DIR and GIT_WORK_TREE early Shroom Moo
  2 siblings, 1 reply; 134+ messages in thread
From: Shroom Moo @ 2026-05-09 13:37 UTC (permalink / raw)
  To: git; +Cc: Johannes Sixt, Mark Levedahl, Aina Boot, Shroom Moo

When git-gui is started inside a .git directory of a non-bare
repository, it should treat the parent directory as the worktree,
as it did before commit 2d92ab32fd (rev-parse: make --show-toplevel
without a worktree an error, 2019-11-19).  However, a bare repository
or a separated gitdir without a worktree must be rejected early.

Protect the previously unguarded calls to `git rev-parse
--show-object-format` and `--show-toplevel`.  Restructure the startup
sequence to:

- Check for a bare repository right after loading the config.  If the
  repository is bare and the current subcommand does not allow bare
  repos (e.g. normal commit mode), show "Cannot use bare repository"
  and exit.

- When `rev-parse --show-toplevel` fails and the repository is
  non-bare, the gitdir path ends with ".git", and we are inside that
  gitdir, use the parent directory as the worktree.  This preserves
  the ability to start git-gui from within a regular repository’s
  .git directory, which was intentionally supported since 87cd09f43e56
  (git-gui: work from the .git dir, 2010-01-23).

- Otherwise, show a descriptive error and exit.

- Wrap `rev-parse --show-object-format` in a catch to avoid a crash
  when the repository configuration is broken (e.g. core.worktree
  pointing to an invalid path).

Also removes the old `_prefix`‑based fallback that computed a relative
path to the worktree top from a subdirectory, and the unconditional
`[file dirname $_gitdir]` guess.  Both are unnecessary now that
`rev‑parse --show‑toplevel` directly provides the absolute top‑level
path and we can `cd` to it.  The guess is further unsafe in
multi‑worktree setups, where a gitdir may have more than one worktree.
The only remaining fallback is the explicit “.git directory” rule for
non‑bare repositories, which mirrors the historical behaviour.
Additionally, only export GIT_WORK_TREE when it is not empty, to avoid
confusing commands in bare-repository subcommands.

This fixes the fatal Tcl error when the working tree is missing, while
keeping the .git startup feature and avoiding any automatic directory
switching that could be dangerous in multi‑worktree setups.

Helped-by: Johannes Sixt <j6t@kdbg•org>
Helped-by: Mark Levedahl <mlevedahl@gmail•com>
Signed-off-by: Shroom Moo <egg_mushroomcow@foxmail•com>
---
 git-gui/git-gui.sh | 76 ++++++++++++++++++++++++++++++----------------
 1 file changed, 49 insertions(+), 27 deletions(-)

diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh
index 23fe76e498..9eb93a76b5 100755
--- a/git-gui/git-gui.sh
+++ b/git-gui/git-gui.sh
@@ -1129,7 +1129,8 @@ if {[catch {
 		}]
 	&& [catch {
 		# beware that from the .git dir this sets _gitdir to .
-		# and _prefix to the empty string
+		# and _prefix to the empty string; this is handled by
+		# the startup safety checks below
 		set _gitdir [git rev-parse --git-dir]
 		set _prefix [git rev-parse --show-prefix]
 	} err]} {
@@ -1142,8 +1143,20 @@ if {[catch {
 	set picked 1
 }
 
+if {![file isdirectory $_gitdir]} {
+	catch {wm withdraw .}
+	error_popup [strcat 
+		[mc "Git directory not found:"] "\n\n$_gitdir\n\n" \
+		[mc "Please ensure GIT_DIR points to a valid Git repository"]]
+	exit 1
+}
+
 # Use object format as hash algorithm (either "sha1" or "sha256")
-set hashalgorithm [git rev-parse --show-object-format]
+if {[catch {set hashalgorithm [git rev-parse --show-object-format]} err]} {
+	catch {wm withdraw .}
+	error_popup [strcat [mc "Failed to determine hash algorithm:"] "\n\n$err"]
+	exit 1
+}
 if {$hashalgorithm eq "sha1"} {
 	set hashlength 40
 } elseif {$hashalgorithm eq "sha256"} {
@@ -1160,46 +1173,52 @@ if {$_gitdir eq "."} {
 	set _gitdir [pwd]
 }
 
-if {![file isdirectory $_gitdir]} {
-	catch {wm withdraw .}
-	error_popup [strcat [mc "Git directory not found:"] "\n\n$_gitdir"]
-	exit 1
-}
 # _gitdir exists, so try loading the config
 load_config 0
 apply_config
 
-set _gitworktree [git rev-parse --show-toplevel]
-
-if {$_prefix ne {}} {
-	if {$_gitworktree eq {}} {
-		regsub -all {[^/]+/} $_prefix ../ cdup
-	} else {
-		set cdup $_gitworktree
-	}
-	if {[catch {cd $cdup} err]} {
+# Handle bare repository and determine working tree
+if {[is_bare]} {
+	# Bare repository: only allowed for certain subcommands
+	if {![is_enabled bare]} {
 		catch {wm withdraw .}
-		error_popup [strcat [mc "Cannot move to top of working directory:"] "\n\n$err"]
+		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n" [file normalize $_gitdir]]
 		exit 1
 	}
-	set _gitworktree [pwd]
-	unset cdup
-} elseif {![is_enabled bare]} {
-	if {[is_bare]} {
-		catch {wm withdraw .}
-		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n$_gitdir"]
-		exit 1
+	# Allowed bare repo does not have a worktree
+	set _gitworktree {}
+} else {
+	# Non-bare repository: we must find a worktree
+	if {[catch {set _gitworktree [git rev-parse --show-toplevel]} err]} {
+		# The only acceptable failure is when we are inside
+		# the .git directory of a regular repository.
+		set inside_gitdir 0
+		catch {set inside_gitdir [git rev-parse --is-inside-git-dir]}
+		if {$inside_gitdir eq {true} && [file tail $_gitdir] eq {.git}} {
+			# Use the parent directory as worktree (historic behavior)
+			set _gitworktree [file normalize [file dirname $_gitdir]]
+		} else {
+			catch {wm withdraw .}
+			error_popup [strcat [mc "Cannot determine working tree:"] "\n\n$err"]
+			exit 1
+		}
 	}
+
 	if {$_gitworktree eq {}} {
-		set _gitworktree [file dirname $_gitdir]
+		catch {wm withdraw .}
+		error_popup [mc "Cannot determine working tree (unexpected empty result)"]
+		exit 1
 	}
+
 	if {[catch {cd $_gitworktree} err]} {
 		catch {wm withdraw .}
-		error_popup [strcat [mc "No working directory"] " $_gitworktree:\n\n$err"]
+		error_popup [strcat [mc "Cannot move to working directory:"] "\n\n$err"]
 		exit 1
 	}
 	set _gitworktree [pwd]
 }
+
+# Derive a human-readable repository name
 set _reponame [file split [file normalize $_gitdir]]
 if {[lindex $_reponame end] eq {.git}} {
 	set _reponame [lindex $_reponame end-1]
@@ -1207,8 +1226,11 @@ if {[lindex $_reponame end] eq {.git}} {
 	set _reponame [lindex $_reponame end]
 }
 
+# Export the final paths
 set env(GIT_DIR) $_gitdir
-set env(GIT_WORK_TREE) $_gitworktree
+if {$_gitworktree ne {}} {
+	set env(GIT_WORK_TREE) $_gitworktree
+}
 
 ######################################################################
 ##
-- 
2.52.0.windows.1


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v7 1/3] git-gui: restructure repository startup
  2026-05-09 13:37               ` [PATCH v7 1/3] git-gui: restructure repository startup Shroom Moo
@ 2026-05-15  8:26                 ` Johannes Sixt
  0 siblings, 0 replies; 134+ messages in thread
From: Johannes Sixt @ 2026-05-15  8:26 UTC (permalink / raw)
  To: Shroom Moo, git; +Cc: Mark Levedahl, Aina Boot

Am 09.05.26 um 15:37 schrieb Shroom Moo:
> When git-gui is started inside a .git directory of a non-bare
> repository, it should treat the parent directory as the worktree,
> as it did before commit 2d92ab32fd (rev-parse: make --show-toplevel
> without a worktree an error, 2019-11-19).  However, a bare repository
> or a separated gitdir without a worktree must be rejected early.
> > Protect the previously unguarded calls to `git rev-parse
> --show-object-format` and `--show-toplevel`.  Restructure the startup
> sequence to:
> 
> - Check for a bare repository right after loading the config.  If the
>   repository is bare and the current subcommand does not allow bare
>   repos (e.g. normal commit mode), show "Cannot use bare repository"
>   and exit.
> 
> - When `rev-parse --show-toplevel` fails and the repository is
>   non-bare, the gitdir path ends with ".git", and we are inside that
>   gitdir, use the parent directory as the worktree.  This preserves
>   the ability to start git-gui from within a regular repository’s
>   .git directory, which was intentionally supported since 87cd09f43e56
>   (git-gui: work from the .git dir, 2010-01-23).
> 
> - Otherwise, show a descriptive error and exit.
> 
> - Wrap `rev-parse --show-object-format` in a catch to avoid a crash
>   when the repository configuration is broken (e.g. core.worktree
>   pointing to an invalid path).
> 
> Also removes the old `_prefix`‑based fallback that computed a relative
> path to the worktree top from a subdirectory, and the unconditional
> `[file dirname $_gitdir]` guess.  Both are unnecessary now that
> `rev‑parse --show‑toplevel` directly provides the absolute top‑level
> path and we can `cd` to it.  The guess is further unsafe in
> multi‑worktree setups, where a gitdir may have more than one worktree.
> The only remaining fallback is the explicit “.git directory” rule for
> non‑bare repositories, which mirrors the historical behaviour.
> Additionally, only export GIT_WORK_TREE when it is not empty, to avoid
> confusing commands in bare-repository subcommands.
> 
> This fixes the fatal Tcl error when the working tree is missing, while
> keeping the .git startup feature and avoiding any automatic directory
> switching that could be dangerous in multi‑worktree setups.

I think that the end result is useful. However, frankly, the patch
attempts to do too many things at once and should still be split further:

- The removal of the cdup fallback could be a preliminary patch.

- The protection of --show-object-format could be a follow-up patch.

> Helped-by: Johannes Sixt <j6t@kdbg•org>
> Helped-by: Mark Levedahl <mlevedahl@gmail•com>
> Signed-off-by: Shroom Moo <egg_mushroomcow@foxmail•com>
> ---
>  git-gui/git-gui.sh | 76 ++++++++++++++++++++++++++++++----------------
>  1 file changed, 49 insertions(+), 27 deletions(-)
> 
> diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh
> index 23fe76e498..9eb93a76b5 100755
> --- a/git-gui/git-gui.sh
> +++ b/git-gui/git-gui.sh
> @@ -1129,7 +1129,8 @@ if {[catch {
>  		}]
>  	&& [catch {
>  		# beware that from the .git dir this sets _gitdir to .
> -		# and _prefix to the empty string
> +		# and _prefix to the empty string; this is handled by
> +		# the startup safety checks below
>  		set _gitdir [git rev-parse --git-dir]
>  		set _prefix [git rev-parse --show-prefix]
>  	} err]} {
> @@ -1142,8 +1143,20 @@ if {[catch {
>  	set picked 1
>  }
>  
> +if {![file isdirectory $_gitdir]} {
> +	catch {wm withdraw .}
> +	error_popup [strcat 
> +		[mc "Git directory not found:"] "\n\n$_gitdir\n\n" \
> +		[mc "Please ensure GIT_DIR points to a valid Git repository"]]
> +	exit 1
> +}
> +

This was moved from below. I would appreciated if there were no changes
in the moved code so that `git diff --color-moved` can show that no
changes were intended. I am not sure that the additional sentence that
mentions GIT_DIR is warranted. If you feel it is needed, please add it
in a separate patch with a justification.

>  # Use object format as hash algorithm (either "sha1" or "sha256")
> -set hashalgorithm [git rev-parse --show-object-format]
> +if {[catch {set hashalgorithm [git rev-parse --show-object-format]} err]} {
> +	catch {wm withdraw .}
> +	error_popup [strcat [mc "Failed to determine hash algorithm:"] "\n\n$err"]
> +	exit 1
> +}
>  if {$hashalgorithm eq "sha1"} {
>  	set hashlength 40
>  } elseif {$hashalgorithm eq "sha256"} {
> @@ -1160,46 +1173,52 @@ if {$_gitdir eq "."} {
>  	set _gitdir [pwd]
>  }
>  
> -if {![file isdirectory $_gitdir]} {
> -	catch {wm withdraw .}
> -	error_popup [strcat [mc "Git directory not found:"] "\n\n$_gitdir"]
> -	exit 1
> -}
>  # _gitdir exists, so try loading the config
>  load_config 0
>  apply_config
>  
> -set _gitworktree [git rev-parse --show-toplevel]
> -
> -if {$_prefix ne {}} {
> -	if {$_gitworktree eq {}} {
> -		regsub -all {[^/]+/} $_prefix ../ cdup
> -	} else {
> -		set cdup $_gitworktree
> -	}
> -	if {[catch {cd $cdup} err]} {
> +# Handle bare repository and determine working tree
> +if {[is_bare]} {
> +	# Bare repository: only allowed for certain subcommands
> +	if {![is_enabled bare]} {
>  		catch {wm withdraw .}
> -		error_popup [strcat [mc "Cannot move to top of working directory:"] "\n\n$err"]
> +		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n" [file normalize $_gitdir]]
>  		exit 1
>  	}
> -	set _gitworktree [pwd]
> -	unset cdup
> -} elseif {![is_enabled bare]} {
> -	if {[is_bare]} {
> -		catch {wm withdraw .}
> -		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n$_gitdir"]
> -		exit 1
> +	# Allowed bare repo does not have a worktree
> +	set _gitworktree {}
> +} else {
> +	# Non-bare repository: we must find a worktree
> +	if {[catch {set _gitworktree [git rev-parse --show-toplevel]} err]} {
> +		# The only acceptable failure is when we are inside
> +		# the .git directory of a regular repository.
> +		set inside_gitdir 0
> +		catch {set inside_gitdir [git rev-parse --is-inside-git-dir]}
> +		if {$inside_gitdir eq {true} && [file tail $_gitdir] eq {.git}} {
> +			# Use the parent directory as worktree (historic behavior)
> +			set _gitworktree [file normalize [file dirname $_gitdir]]
> +		} else {
> +			catch {wm withdraw .}
> +			error_popup [strcat [mc "Cannot determine working tree:"] "\n\n$err"]
> +			exit 1
> +		}
>  	}
> +
>  	if {$_gitworktree eq {}} {
> -		set _gitworktree [file dirname $_gitdir]
> +		catch {wm withdraw .}
> +		error_popup [mc "Cannot determine working tree (unexpected empty result)"]
> +		exit 1
>  	}

An empty $_gitworktree should be practically impossible at this point.
Personally, I would let the following "cd" handle the case (it fails if
the argument is empty).

> +
>  	if {[catch {cd $_gitworktree} err]} {
>  		catch {wm withdraw .}
> -		error_popup [strcat [mc "No working directory"] " $_gitworktree:\n\n$err"]
> +		error_popup [strcat [mc "Cannot move to working directory:"] "\n\n$err"]
>  		exit 1
>  	}
>  	set _gitworktree [pwd]
>  }
> +
> +# Derive a human-readable repository name
>  set _reponame [file split [file normalize $_gitdir]]
>  if {[lindex $_reponame end] eq {.git}} {
>  	set _reponame [lindex $_reponame end-1]
> @@ -1207,8 +1226,11 @@ if {[lindex $_reponame end] eq {.git}} {
>  	set _reponame [lindex $_reponame end]
>  }
>  
> +# Export the final paths
>  set env(GIT_DIR) $_gitdir
> -set env(GIT_WORK_TREE) $_gitworktree
> +if {$_gitworktree ne {}} {
> +	set env(GIT_WORK_TREE) $_gitworktree
> +}
>  
>  ######################################################################
>  ##

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v7 2/3] git-gui: disable gitk visualization when no worktree available
       [not found]             ` <20260509133756.1367-1-egg_mushroomcow@foxmail.com>
  2026-05-09 13:37               ` [PATCH v7 1/3] git-gui: restructure repository startup Shroom Moo
@ 2026-05-09 13:37               ` Shroom Moo
  2026-05-15  8:28                 ` Johannes Sixt
  2026-05-09 13:37               ` [PATCH v7 3/3] git-gui: handle GIT_DIR and GIT_WORK_TREE early Shroom Moo
  2 siblings, 1 reply; 134+ messages in thread
From: Shroom Moo @ 2026-05-09 13:37 UTC (permalink / raw)
  To: git; +Cc: Johannes Sixt, Mark Levedahl, Aina Boot, Shroom Moo

When git-gui is started in a bare repository with the 'bare' option
enabled (e.g., for blame/browser), there is no working tree.  The
"Visualize Current Branch's History" and "Visualize All Branch
History" menu items remain enabled, but clicking them triggers a Tcl
error because do_gitk tries to change directory to an empty
_gitworktree.

Fix this by disabling the two visualization menu items when the
repository is bare and the 'bare' option is active.  Also update
current_branch_write to keep the state consistent when the branch
changes, and add a defensive check in do_gitk to avoid the error
should the menu state somehow become out of sync.

This complements the startup sequence improvements in the previous
commit, which already correctly identifies bare repositories and
leaves _gitworktree empty in such cases.

Helped-by: Mark Levedahl <mlevedahl@gmail•com>
Helped-by: Johannes Sixt <j6t@kdbg•org>
Signed-off-by: Shroom Moo <egg_mushroomcow@foxmail•com>
---
 git-gui/git-gui.sh | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh
index 9eb93a76b5..ff022c5bbb 100755
--- a/git-gui/git-gui.sh
+++ b/git-gui/git-gui.sh
@@ -2036,6 +2036,10 @@ proc do_gitk {revs {is_submodule false}} {
 	#
 	set exe [_which gitk -script]
 	set cmd [list [info nameofexecutable] $exe]
+	if {$_gitworktree eq {}} {
+		error_popup [mc "Cannot visualize history: no working tree"]
+		return
+	}
 	if {$exe eq {}} {
 		error_popup [mc "Couldn't find gitk in PATH"]
 	} else {
@@ -2659,6 +2663,13 @@ set ui_visualize_current [.mbar.repository index last]
 .mbar.repository add command \
 	-label [mc "Visualize All Branch History"] \
 	-command {do_gitk --all}
+set ui_visualize_all [.mbar.repository index last]
+
+# Cannot work without a working tree
+if {[is_bare] && [is_enabled bare]} {
+	.mbar.repository entryconf $ui_visualize_current -state disabled
+	.mbar.repository entryconf $ui_visualize_all   -state disabled
+}
 .mbar.repository add separator
 
 proc current_branch_write {args} {
@@ -2667,6 +2678,13 @@ proc current_branch_write {args} {
 		-label [mc "Browse %s's Files" $current_branch]
 	.mbar.repository entryconf $::ui_visualize_current \
 		-label [mc "Visualize %s's History" $current_branch]
+	if {[is_bare] && [is_enabled bare]} {
+		.mbar.repository entryconf $::ui_visualize_current -state disabled
+		.mbar.repository entryconf $::ui_visualize_all   -state disabled
+	} else {
+		.mbar.repository entryconf $::ui_visualize_current -state normal
+		.mbar.repository entryconf $::ui_visualize_all   -state normal
+	}
 }
 trace add variable current_branch write current_branch_write
 
-- 
2.52.0.windows.1



^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v7 2/3] git-gui: disable gitk visualization when no worktree available
  2026-05-09 13:37               ` [PATCH v7 2/3] git-gui: disable gitk visualization when no worktree available Shroom Moo
@ 2026-05-15  8:28                 ` Johannes Sixt
  0 siblings, 0 replies; 134+ messages in thread
From: Johannes Sixt @ 2026-05-15  8:28 UTC (permalink / raw)
  To: Shroom Moo, git; +Cc: Mark Levedahl, Aina Boot

Am 09.05.26 um 15:37 schrieb Shroom Moo:
> When git-gui is started in a bare repository with the 'bare' option
> enabled (e.g., for blame/browser), there is no working tree.  The
> "Visualize Current Branch's History" and "Visualize All Branch
> History" menu items remain enabled, but clicking them triggers a Tcl
> error because do_gitk tries to change directory to an empty
> _gitworktree.

I cannot reproduce this claim. The failure is not a Tcl error, but an
error in some `git` invocation that cannot handle an empty
GIT_WORK_TREE. And that happens only beginning with the *second*
invocation of one of the "Visualize" calls, because then an empty
GIT_WORK_TREE is exported into the environment.

> 
> Fix this by disabling the two visualization menu items when the
> repository is bare and the 'bare' option is active.  Also update
> current_branch_write to keep the state consistent when the branch
> changes, and add a defensive check in do_gitk to avoid the error
> should the menu state somehow become out of sync.

This change is not correct. Gitk can operate without a working tree. The
menu entries should not be disabled, ever. The bug is somewhere else.

See also my suggested replacement patch in my reply to 3/3.

-- Hannes


^ permalink raw reply	[flat|nested] 134+ messages in thread

* [PATCH v7 3/3] git-gui: handle GIT_DIR and GIT_WORK_TREE early
       [not found]             ` <20260509133756.1367-1-egg_mushroomcow@foxmail.com>
  2026-05-09 13:37               ` [PATCH v7 1/3] git-gui: restructure repository startup Shroom Moo
  2026-05-09 13:37               ` [PATCH v7 2/3] git-gui: disable gitk visualization when no worktree available Shroom Moo
@ 2026-05-09 13:37               ` Shroom Moo
  2026-05-15  8:28                 ` Johannes Sixt
  2 siblings, 1 reply; 134+ messages in thread
From: Shroom Moo @ 2026-05-09 13:37 UTC (permalink / raw)
  To: git; +Cc: Johannes Sixt, Mark Levedahl, Aina Boot, Shroom Moo

Users expect these two invocations to be equivalent:

    GIT_WORK_TREE=/some/path GIT_DIR=/some/path/.git git gui
    git -C /some/path gui

Currently, the environment variable variant often brings up the
repository picker or ignores the requested worktree because
GIT_WORK_TREE is processed too late.  Moreover, after determining
the working tree, git-gui unconditionally exports GIT_WORK_TREE.
When no worktree is found (e.g., in a bare repository with a
read-only subcommand like blame), an empty value is exported, which
confuses commands like `git branch --show-current`.

Fix both issues:

- Introduce resolve_initial_environment, which is called before any
  repository detection.  When both GIT_DIR and GIT_WORK_TREE are
  set, it changes to the specified worktree, verifies that it is a
  valid worktree, and adopts the given GIT_DIR.  In this case, the
  subsequent automatic detection is skipped.  After successful
  validation, GIT_WORK_TREE is unset so that later steps do not
  inherit it.  Export the absolute GIT_DIR before changing directory
  to ensure Git commands see the intended repository.

- Introduce set_worktree_env and unset_worktree_env helper procedures
  to safely manage the GIT_WORK_TREE environment variable:
  set_worktree_env sets it only when $_gitworktree is non-empty,
  and unset_worktree_env removes the variable without error.
  Replace all direct manipulations of GIT_WORK_TREE with these
  helpers to avoid accidentally exporting an empty value or
  causing 'unset' errors.

- At the end of startup, use set_worktree_env instead of exporting
  GIT_WORK_TREE unconditionally.  This leaves the variable unset
  when no worktree exists (bare repository), fixing commands that
  would fail with an empty GIT_WORK_TREE.

When only GIT_DIR is set without GIT_WORK_TREE, the existing logic
remains unchanged for backward compatibility.  Setting only
GIT_WORK_TREE is not a valid Git use-case and is silently ignored.

Helped-by: Mark Levedahl <mlevedahl@gmail•com>
Helped-by: Johannes Sixt <j6t@kdbg•org>
Signed-off-by: Shroom Moo <egg_mushroomcow@foxmail•com>
---
 git-gui/git-gui.sh | 103 ++++++++++++++++++++++++++++++++++-----------
 1 file changed, 79 insertions(+), 24 deletions(-)

diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh
index ff022c5bbb..1123f9855b 100755
--- a/git-gui/git-gui.sh
+++ b/git-gui/git-gui.sh
@@ -1122,25 +1122,82 @@ unset argv0dir
 ##
 ## repository setup
 
+# Safely set/unset GIT_WORK_TREE
+proc set_worktree_env {} {
+    global _gitworktree
+    if {$_gitworktree ne {}} {
+        set ::env(GIT_WORK_TREE) $_gitworktree
+    }
+}
+
+proc unset_worktree_env {} {
+    catch {unset ::env(GIT_WORK_TREE)}
+}
+
+proc resolve_initial_environment {} {
+	global _gitdir env
+
+	# Only act if both GIT_DIR and GIT_WORK_TREE are set
+	if {[info exists env(GIT_DIR)] && [info exists env(GIT_WORK_TREE)]} {
+		# Validate GIT_DIR by resolving its absolute path
+		if {[catch {set _abs [git rev-parse --absolute-git-dir]} err]} {
+			catch {wm withdraw .}
+			error_popup [strcat [mc "Invalid GIT_DIR:"] "\n\n$err"]
+			exit 1
+		}
+		set env(GIT_DIR) $_abs
+
+		# Change current directory to GIT_WORK_TREE
+		if {[catch {cd $env(GIT_WORK_TREE)} err]} {
+			catch {wm withdraw .}
+			error_popup [strcat [mc "Cannot change to GIT_WORK_TREE:"] "\n\n$err"]
+			exit 1
+		}
+
+		# Verify that GIT_WORK_TREE is a valid Git worktree
+		if {[catch {git rev-parse --show-toplevel} err]} {
+			catch {wm withdraw .}
+			error_popup [strcat [mc "GIT_WORK_TREE is not a valid worktree:"] "\n\n$err"]
+			exit 1
+		}
+
+		# Use repository path specified by environment variables
+		set _gitdir $_abs
+		set ::_prefix {}
+
+		# Unset GIT_WORK_TREE to prevent it from being inherited by child processes
+		unset_worktree_env
+
+		return 1
+	}
+
+	# Other cases (only GIT_DIR, only GIT_WORK_TREE, or neither):
+	# Do nothing here and let the existing logic handle it later
+	return 0
+}
+set force_gitdir [resolve_initial_environment]
+
 set picked 0
-if {[catch {
-		set _gitdir $env(GIT_DIR)
-		set _prefix {}
-		}]
-	&& [catch {
-		# beware that from the .git dir this sets _gitdir to .
-		# and _prefix to the empty string; this is handled by
-		# the startup safety checks below
-		set _gitdir [git rev-parse --git-dir]
-		set _prefix [git rev-parse --show-prefix]
-	} err]} {
-	load_config 1
-	apply_config
-	choose_repository::pick
-	if {![file isdirectory $_gitdir]} {
-		exit 1
+if {!$force_gitdir} {
+	if {[catch {
+			set _gitdir $env(GIT_DIR)
+			set _prefix {}
+			}]
+		&& [catch {
+			# beware that from the .git dir this sets _gitdir to .
+			# and _prefix to the empty string; this is handled by
+			# the startup safety checks below
+			set _gitdir [git rev-parse --git-dir]
+			set _prefix [git rev-parse --show-prefix]
+		} err]} {
+		load_config 1
+		apply_config
+		choose_repository::pick
+		if {![file isdirectory $_gitdir]} {
+			exit 1
+		}
+		set picked 1
 	}
-	set picked 1
 }
 
 if {![file isdirectory $_gitdir]} {
@@ -1228,9 +1285,7 @@ if {[lindex $_reponame end] eq {.git}} {
 
 # Export the final paths
 set env(GIT_DIR) $_gitdir
-if {$_gitworktree ne {}} {
-	set env(GIT_WORK_TREE) $_gitworktree
-}
+set_worktree_env
 
 ######################################################################
 ##
@@ -2077,12 +2132,12 @@ proc do_gitk {revs {is_submodule false}} {
 			# by setting these to the appropriate values to allow gitk
 			# to skip the heuristics to find their proper value
 			unset env(GIT_DIR)
-			unset env(GIT_WORK_TREE)
+			unset_worktree_env
 		}
 		safe_exec_bg [concat $cmd $revs "--" "--"]
 
 		set env(GIT_DIR) $_gitdir
-		set env(GIT_WORK_TREE) $_gitworktree
+		set_worktree_env
 		cd $pwd
 
 		if {[info exists main_status]} {
@@ -2111,7 +2166,7 @@ proc do_git_gui {} {
 		# see note in do_gitk about unsetting these vars when
 		# running tools in a submodule
 		unset env(GIT_DIR)
-		unset env(GIT_WORK_TREE)
+		unset_worktree_env
 
 		set pwd [pwd]
 		cd $current_diff_path
@@ -2119,7 +2174,7 @@ proc do_git_gui {} {
 		safe_exec_bg [concat $exe gui]
 
 		set env(GIT_DIR) $_gitdir
-		set env(GIT_WORK_TREE) $_gitworktree
+		set_worktree_env
 		cd $pwd
 
 		set status_operation [$::main_status \
-- 
2.52.0.windows.1


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v7 3/3] git-gui: handle GIT_DIR and GIT_WORK_TREE early
  2026-05-09 13:37               ` [PATCH v7 3/3] git-gui: handle GIT_DIR and GIT_WORK_TREE early Shroom Moo
@ 2026-05-15  8:28                 ` Johannes Sixt
  0 siblings, 0 replies; 134+ messages in thread
From: Johannes Sixt @ 2026-05-15  8:28 UTC (permalink / raw)
  To: Shroom Moo, git; +Cc: Mark Levedahl, Aina Boot

Am 09.05.26 um 15:37 schrieb Shroom Moo:
> Users expect these two invocations to be equivalent:
> 
>     GIT_WORK_TREE=/some/path GIT_DIR=/some/path/.git git gui
>     git -C /some/path gui
> 
> Currently, the environment variable variant often brings up the
> repository picker or ignores the requested worktree because
> GIT_WORK_TREE is processed too late.

I cannot reproduce the case that brings the repository picker. All other
failure cases that I can produce are reasonable and do not indicate that
GIT_WORK_TREE is processed too late.

>  Moreover, after determining
> the working tree, git-gui unconditionally exports GIT_WORK_TREE.
> When no worktree is found (e.g., in a bare repository with a
> read-only subcommand like blame), an empty value is exported, which
> confuses commands like `git branch --show-current`.

True. I think the culprit is that we export GIT_WORK_TREE in the first
place.

I suggest the following patch to replace this and the previous patch.

---- 8< ----
From: Johannes Sixt <j6t@kdbg•org>
Subject: [PATCH] git-gui: operate git commands without GIT_WORK_TREE

The manual page of the git command states about the --git-dir option:

   Specifying the location of the ".git" directory using this option
   (or GIT_DIR environment variable) turns off the repository
   discovery [...], and tells Git that you are at the top level of
   the working tree.

Use this to our advantage:

- Set GIT_DIR in the environment to the value that was discovered, so
  that the invoked git commands operate on the same repository
  database that Git GUI uses even after it changes the working
  directory.

- After changing the working directory to the top level of the working
  tree, ensure that GIT_WORK_TREE is not set, because, as per
  documentation, all git invocations from then on will assume that the
  current working directory is also the top level working tree.

- Remove the now obsolete GIT_WORK_TREE dance when subordinate Gitk or
  Git GUI are invoked for a submodule.

Do keep the state of GIT_WORK_TREE if we are in a bare repository,
because Git GUI is not interested in the worktree at all, as no commit
mode is possible in a bare repository.

This avoids cases where an empty GIT_WORK_TREE was exported into the
environment, most notably by a call of `git gui blame HEAD file` in a
bare repository.

Signed-off-by: Johannes Sixt <j6t@kdbg•org>
---
 git-gui.sh | 24 +++++++-----------------
 1 file changed, 7 insertions(+), 17 deletions(-)

diff --git a/git-gui.sh b/git-gui.sh
index 76560ec825cf..146a29a809a8 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -1216,6 +1216,7 @@ if {[is_bare]} {
 		exit 1
 	}
 	set _gitworktree [pwd]
+	catch { unset env(GIT_WORK_TREE) }
 }
 
 # Derive a human-readable repository name
@@ -1228,9 +1229,6 @@ if {[lindex $_reponame end] eq {.git}} {
 
 # Export the final paths
 set env(GIT_DIR) $_gitdir
-if {$_gitworktree ne {}} {
-	set env(GIT_WORK_TREE) $_gitworktree
-}
 
 ######################################################################
 ##
@@ -2029,7 +2027,7 @@ proc incr_font_size {font {amt 1}} {
 
 proc do_gitk {revs {is_submodule false}} {
 	global current_diff_path file_states current_diff_side ui_index
-	global _gitdir _gitworktree
+	global _gitdir
 
 	# -- Always start gitk through whatever we were loaded with.  This
 	#    lets us bypass using shell process on Windows systems.
@@ -2043,11 +2041,7 @@ proc do_gitk {revs {is_submodule false}} {
 
 		set pwd [pwd]
 
-		if {!$is_submodule} {
-			if {![is_bare]} {
-				cd $_gitworktree
-			}
-		} else {
+		if {$is_submodule} {
 			cd $current_diff_path
 			if {$revs eq {--}} {
 				set s $file_states($current_diff_path)
@@ -2067,18 +2061,16 @@ proc do_gitk {revs {is_submodule false}} {
 				}
 				set revs $old_sha1...$new_sha1
 			}
-			# GIT_DIR and GIT_WORK_TREE for the submodule are not the ones
-			# we've been using for the main repository, so unset them.
+			# GIT_DIR for the submodule is not the one we've been using for
+			# the main repository, so unset it. (GIT_WORK_TREE is already unset.)
 			# TODO we could make life easier (start up faster?) for gitk
 			# by setting these to the appropriate values to allow gitk
 			# to skip the heuristics to find their proper value
 			unset env(GIT_DIR)
-			unset env(GIT_WORK_TREE)
 		}
 		safe_exec_bg [concat $cmd $revs "--" "--"]
 
 		set env(GIT_DIR) $_gitdir
-		set env(GIT_WORK_TREE) $_gitworktree
 		cd $pwd
 
 		if {[info exists main_status]} {
@@ -2102,12 +2094,11 @@ proc do_git_gui {} {
 		error_popup [mc "Couldn't find git gui in PATH"]
 	} else {
 		global env
-		global _gitdir _gitworktree
+		global _gitdir
 
-		# see note in do_gitk about unsetting these vars when
+		# see note in do_gitk about unsetting this variable when
 		# running tools in a submodule
 		unset env(GIT_DIR)
-		unset env(GIT_WORK_TREE)
 
 		set pwd [pwd]
 		cd $current_diff_path
@@ -2115,7 +2106,6 @@ proc do_git_gui {} {
 		safe_exec_bg [concat $exe gui]
 
 		set env(GIT_DIR) $_gitdir
-		set env(GIT_WORK_TREE) $_gitworktree
 		cd $pwd
 
 		set status_operation [$::main_status \
-- 
2.54.0.215.g4fe990ec16


^ permalink raw reply related	[flat|nested] 134+ messages in thread

[parent not found: <20260506202751.3294-1-egg_mushroomcow@foxmail.com>]

* [PATCH v6 1/3] git-gui: restructure repository startup
       [not found]           ` <20260506202751.3294-1-egg_mushroomcow@foxmail.com>
@ 2026-05-06 20:27             ` Shroom Moo
  2026-05-06 20:27             ` [PATCH v6 2/3] git-gui: disable gitk visualization when no worktree available Shroom Moo
  2026-05-06 20:27             ` [PATCH v6 3/3] git-gui: handle GIT_DIR and GIT_WORK_TREE early Shroom Moo
  2 siblings, 0 replies; 134+ messages in thread
From: Shroom Moo @ 2026-05-06 20:27 UTC (permalink / raw)
  To: git; +Cc: Johannes Sixt, Mark Levedahl, Shroom Moo

When git-gui is started inside a .git directory of a non-bare
repository, it should treat the parent directory as the worktree,
as it did before commit 2d92ab32fd (rev-parse: make --show-toplevel
without a worktree an error, 2019-11-19).  However, a bare repository
or a separated gitdir without a worktree must be rejected early.

Protect the previously unguarded calls to `git rev-parse
--show-object-format` and `--show-toplevel`.  Restructure the startup
sequence to:

- Check for a bare repository right after loading the config.  If the
  repository is bare and the current subcommand does not allow bare
  repos (e.g. normal commit mode), show "Cannot use bare repository"
  and exit.

- When `rev-parse --show-toplevel` fails and the repository is
  non-bare, the gitdir path ends with ".git", and we are inside that
  gitdir, use the parent directory as the worktree.  This preserves
  the ability to start git-gui from within a regular repository’s
  .git directory, which was intentionally supported since 87cd09f43e56
  (git-gui: work from the .git dir, 2010-01-23).

- Otherwise, show a descriptive error and exit.

- Wrap `rev-parse --show-object-format` in a catch to avoid a crash
  when the repository configuration is broken (e.g. core.worktree
  pointing to an invalid path).

Also removes the old `_prefix`‑based fallback that computed a relative
path to the worktree top from a subdirectory, and the unconditional
`[file dirname $_gitdir]` guess.  Both are unnecessary now that
`rev‑parse --show‑toplevel` directly provides the absolute top‑level
path and we can `cd` to it.  The guess is further unsafe in
multi‑worktree setups, where a gitdir may have more than one worktree.
The only remaining fallback is the explicit “.git directory” rule for
non‑bare repositories, which mirrors the historical behaviour.

This fixes the fatal Tcl error when the working tree is missing, while
keeping the .git startup feature and avoiding any automatic directory
switching that could be dangerous in multi‑worktree setups.

Helped-by: Johannes Sixt <j6t@kdbg•org>
Helped-by: Mark Levedahl <mlevedahl@gmail•com>
Signed-off-by: Shroom Moo <egg_mushroomcow@foxmail•com>
---
 git-gui/git-gui.sh | 72 +++++++++++++++++++++++++++++-----------------
 1 file changed, 46 insertions(+), 26 deletions(-)

diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh
index 23fe76e498..fbdc0b2a41 100755
--- a/git-gui/git-gui.sh
+++ b/git-gui/git-gui.sh
@@ -1129,7 +1129,8 @@ if {[catch {
 		}]
 	&& [catch {
 		# beware that from the .git dir this sets _gitdir to .
-		# and _prefix to the empty string
+		# and _prefix to the empty string; this is handled by
+		# the startup safety checks below
 		set _gitdir [git rev-parse --git-dir]
 		set _prefix [git rev-parse --show-prefix]
 	} err]} {
@@ -1142,8 +1143,20 @@ if {[catch {
 	set picked 1
 }
 
+if {![file isdirectory $_gitdir]} {
+	catch {wm withdraw .}
+	error_popup [strcat 
+		[mc "Git directory not found:"] "\n\n$_gitdir\n\n" \
+		[mc "Please ensure GIT_DIR points to a valid Git repository"]]
+	exit 1
+}
+
 # Use object format as hash algorithm (either "sha1" or "sha256")
-set hashalgorithm [git rev-parse --show-object-format]
+if {[catch {set hashalgorithm [git rev-parse --show-object-format]} err]} {
+	catch {wm withdraw .}
+	error_popup [strcat [mc "Failed to determine hash algorithm:"] "\n\n$err"]
+	exit 1
+}
 if {$hashalgorithm eq "sha1"} {
 	set hashlength 40
 } elseif {$hashalgorithm eq "sha256"} {
@@ -1160,46 +1173,52 @@ if {$_gitdir eq "."} {
 	set _gitdir [pwd]
 }
 
-if {![file isdirectory $_gitdir]} {
-	catch {wm withdraw .}
-	error_popup [strcat [mc "Git directory not found:"] "\n\n$_gitdir"]
-	exit 1
-}
 # _gitdir exists, so try loading the config
 load_config 0
 apply_config
 
-set _gitworktree [git rev-parse --show-toplevel]
-
-if {$_prefix ne {}} {
-	if {$_gitworktree eq {}} {
-		regsub -all {[^/]+/} $_prefix ../ cdup
-	} else {
-		set cdup $_gitworktree
-	}
-	if {[catch {cd $cdup} err]} {
+# Handle bare repository and determine working tree
+if {[is_bare]} {
+	# Bare repository: only allowed for certain subcommands
+	if {![is_enabled bare]} {
 		catch {wm withdraw .}
-		error_popup [strcat [mc "Cannot move to top of working directory:"] "\n\n$err"]
+		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n" [file normalize $_gitdir]]
 		exit 1
 	}
-	set _gitworktree [pwd]
-	unset cdup
-} elseif {![is_enabled bare]} {
-	if {[is_bare]} {
-		catch {wm withdraw .}
-		error_popup [strcat [mc "Cannot use bare repository:"] "\n\n$_gitdir"]
-		exit 1
+	# Allowed bare repo does not have a worktree
+	set _gitworktree {}
+} else {
+	# Non-bare repository: we must find a worktree
+	if {[catch {set _gitworktree [git rev-parse --show-toplevel]} err]} {
+		# The only acceptable failure is when we are inside
+		# the .git directory of a regular repository.
+		set inside_gitdir 0
+		catch {set inside_gitdir [git rev-parse --is-inside-git-dir]}
+		if {$inside_gitdir eq {true} && [file tail $_gitdir] eq {.git}} {
+			# Use the parent directory as worktree (historic behavior)
+			set _gitworktree [file normalize [file dirname $_gitdir]]
+		} else {
+			catch {wm withdraw .}
+			error_popup [strcat [mc "Cannot determine working tree:"] "\n\n$err"]
+			exit 1
+		}
 	}
+
 	if {$_gitworktree eq {}} {
-		set _gitworktree [file dirname $_gitdir]
+		catch {wm withdraw .}
+		error_popup [mc "Cannot determine working tree (unexpected empty result)"]
+		exit 1
 	}
+
 	if {[catch {cd $_gitworktree} err]} {
 		catch {wm withdraw .}
-		error_popup [strcat [mc "No working directory"] " $_gitworktree:\n\n$err"]
+		error_popup [strcat [mc "Cannot move to working directory:"] "\n\n$err"]
 		exit 1
 	}
 	set _gitworktree [pwd]
 }
+
+# Derive a human-readable repository name
 set _reponame [file split [file normalize $_gitdir]]
 if {[lindex $_reponame end] eq {.git}} {
 	set _reponame [lindex $_reponame end-1]
@@ -1207,6 +1226,7 @@ if {[lindex $_reponame end] eq {.git}} {
 	set _reponame [lindex $_reponame end]
 }
 
+# Export the final paths
 set env(GIT_DIR) $_gitdir
 set env(GIT_WORK_TREE) $_gitworktree
 
-- 
2.52.0.windows.1


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* [PATCH v6 2/3] git-gui: disable gitk visualization when no worktree available
       [not found]           ` <20260506202751.3294-1-egg_mushroomcow@foxmail.com>
  2026-05-06 20:27             ` [PATCH v6 1/3] git-gui: restructure repository startup Shroom Moo
@ 2026-05-06 20:27             ` Shroom Moo
  2026-05-06 20:27             ` [PATCH v6 3/3] git-gui: handle GIT_DIR and GIT_WORK_TREE early Shroom Moo
  2 siblings, 0 replies; 134+ messages in thread
From: Shroom Moo @ 2026-05-06 20:27 UTC (permalink / raw)
  To: git; +Cc: Johannes Sixt, Mark Levedahl, Shroom Moo

When git-gui is started in a bare repository with the 'bare' option
enabled (e.g., for blame/browser), there is no working tree.  The
"Visualize Current Branch's History" and "Visualize All Branch
History" menu items remain enabled, but clicking them triggers a Tcl
error because do_gitk tries to change directory to an empty
_gitworktree.

Fix this by disabling the two visualization menu items when the
repository is bare and the 'bare' option is active.  Also update
current_branch_write to keep the state consistent when the branch
changes, and add a defensive check in do_gitk to avoid the error
should the menu state somehow become out of sync.

This complements the startup sequence improvements in the previous
commit, which already correctly identifies bare repositories and
leaves _gitworktree empty in such cases.

Helped-by: Mark Levedahl <mlevedahl@gmail•com>
Helped-by: Johannes Sixt <j6t@kdbg•org>
Signed-off-by: Shroom Moo <egg_mushroomcow@foxmail•com>
---
 git-gui/git-gui.sh | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh
index fbdc0b2a41..1191e6654c 100755
--- a/git-gui/git-gui.sh
+++ b/git-gui/git-gui.sh
@@ -2034,6 +2034,10 @@ proc do_gitk {revs {is_submodule false}} {
 	#
 	set exe [_which gitk -script]
 	set cmd [list [info nameofexecutable] $exe]
+	if {$_gitworktree eq {}} {
+		error_popup [mc "Cannot visualize history: no working tree"]
+		return
+	}
 	if {$exe eq {}} {
 		error_popup [mc "Couldn't find gitk in PATH"]
 	} else {
@@ -2657,6 +2661,13 @@ set ui_visualize_current [.mbar.repository index last]
 .mbar.repository add command \
 	-label [mc "Visualize All Branch History"] \
 	-command {do_gitk --all}
+set ui_visualize_all [.mbar.repository index last]
+
+# Cannot work without a working tree
+if {[is_bare] && [is_enabled bare]} {
+	.mbar.repository entryconf $ui_visualize_current -state disabled
+	.mbar.repository entryconf $ui_visualize_all   -state disabled
+}
 .mbar.repository add separator
 
 proc current_branch_write {args} {
@@ -2665,6 +2676,13 @@ proc current_branch_write {args} {
 		-label [mc "Browse %s's Files" $current_branch]
 	.mbar.repository entryconf $::ui_visualize_current \
 		-label [mc "Visualize %s's History" $current_branch]
+	if {[is_bare] && [is_enabled bare]} {
+		.mbar.repository entryconf $::ui_visualize_current -state disabled
+		.mbar.repository entryconf $::ui_visualize_all   -state disabled
+	} else {
+		.mbar.repository entryconf $::ui_visualize_current -state normal
+		.mbar.repository entryconf $::ui_visualize_all   -state normal
+	}
 }
 trace add variable current_branch write current_branch_write
 
-- 
2.52.0.windows.1



^ permalink raw reply related	[flat|nested] 134+ messages in thread

* [PATCH v6 3/3] git-gui: handle GIT_DIR and GIT_WORK_TREE early
       [not found]           ` <20260506202751.3294-1-egg_mushroomcow@foxmail.com>
  2026-05-06 20:27             ` [PATCH v6 1/3] git-gui: restructure repository startup Shroom Moo
  2026-05-06 20:27             ` [PATCH v6 2/3] git-gui: disable gitk visualization when no worktree available Shroom Moo
@ 2026-05-06 20:27             ` Shroom Moo
  2026-05-07 15:50               ` Mark Levedahl
  2 siblings, 1 reply; 134+ messages in thread
From: Shroom Moo @ 2026-05-06 20:27 UTC (permalink / raw)
  To: git; +Cc: Johannes Sixt, Mark Levedahl, Shroom Moo

Users expect these two invocations to be equivalent:

    GIT_WORK_TREE=/some/path GIT_DIR=/some/path/.git git gui
    git -C /some/path gui

Currently, the environment variable variant often brings up the
repository picker or ignores the requested worktree because
GIT_WORK_TREE is processed too late.  Moreover, after determining
the working tree, git-gui unconditionally exports GIT_WORK_TREE.
When no worktree is found (e.g., in a bare repository with a
read-only subcommand like blame), an empty value is exported, which
confuses commands like `git branch --show-current`.

Fix both issues:

- Introduce resolve_initial_environment, which is called before any
  repository detection.  When both GIT_DIR and GIT_WORK_TREE are
  set, it changes to the specified worktree, verifies that it is a
  valid worktree, and adopts the given GIT_DIR.  In this case, the
  subsequent automatic detection is skipped.  After successful
  validation, GIT_WORK_TREE is unset so that later steps do not
  inherit it.

- At the end of startup, only export GIT_DIR.  The current directory
  is already at the top of the worktree (when a worktree exists), so
  Git can discover the worktree automatically.  Not exporting an
  empty GIT_WORK_TREE fixes `git gui blame` in bare repositories.

When only GIT_DIR is set without GIT_WORK_TREE, the existing logic
remains unchanged for backward compatibility.  Setting only
GIT_WORK_TREE is not a valid Git use-case and is silently ignored.

Helped-by: Mark Levedahl <mlevedahl@gmail•com>
Helped-by: Johannes Sixt <j6t@kdbg•org>
Signed-off-by: Shroom Moo <egg_mushroomcow@foxmail•com>
---
 git-gui/git-gui.sh | 84 ++++++++++++++++++++++++++++++++++++----------
 1 file changed, 66 insertions(+), 18 deletions(-)

diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh
index 1191e6654c..ca651dc576 100755
--- a/git-gui/git-gui.sh
+++ b/git-gui/git-gui.sh
@@ -1122,25 +1122,69 @@ unset argv0dir
 ##
 ## repository setup
 
+proc resolve_initial_environment {} {
+	global _gitdir env
+
+	# Only act if both GIT_DIR and GIT_WORK_TREE are set
+	if {[info exists env(GIT_DIR)] && [info exists env(GIT_WORK_TREE)]} {
+		# Validate GIT_DIR by resolving its absolute path
+		if {[catch {set _abs [git rev-parse --absolute-git-dir]} err]} {
+			catch {wm withdraw .}
+			error_popup [strcat [mc "Invalid GIT_DIR:"] "\n\n$err"]
+			exit 1
+		}
+
+		# Change current directory to GIT_WORK_TREE
+		if {[catch {cd $env(GIT_WORK_TREE)} err]} {
+			catch {wm withdraw .}
+			error_popup [strcat [mc "Cannot change to GIT_WORK_TREE:"] "\n\n$err"]
+			exit 1
+		}
+
+		# Verify that GIT_WORK_TREE is a valid Git worktree
+		if {[catch {git rev-parse --show-toplevel} err]} {
+			catch {wm withdraw .}
+			error_popup [strcat [mc "GIT_WORK_TREE is not a valid worktree:"] "\n\n$err"]
+			exit 1
+		}
+
+		# Use repository path specified by environment variables
+		set _gitdir $_abs
+		set ::_prefix {}
+
+		# Unset GIT_WORK_TREE to prevent it from being inherited by child processes
+		unset env(GIT_WORK_TREE)
+
+		return 1
+	}
+
+	# Other cases (only GIT_DIR, only GIT_WORK_TREE, or neither):
+	# Do nothing here and let the existing logic handle it later
+	return 0
+}
+set force_gitdir [resolve_initial_environment]
+
 set picked 0
-if {[catch {
-		set _gitdir $env(GIT_DIR)
-		set _prefix {}
-		}]
-	&& [catch {
-		# beware that from the .git dir this sets _gitdir to .
-		# and _prefix to the empty string; this is handled by
-		# the startup safety checks below
-		set _gitdir [git rev-parse --git-dir]
-		set _prefix [git rev-parse --show-prefix]
-	} err]} {
-	load_config 1
-	apply_config
-	choose_repository::pick
-	if {![file isdirectory $_gitdir]} {
-		exit 1
+if {!$force_gitdir} {
+	if {[catch {
+			set _gitdir $env(GIT_DIR)
+			set _prefix {}
+			}]
+		&& [catch {
+			# beware that from the .git dir this sets _gitdir to .
+			# and _prefix to the empty string; this is handled by
+			# the startup safety checks below
+			set _gitdir [git rev-parse --git-dir]
+			set _prefix [git rev-parse --show-prefix]
+		} err]} {
+		load_config 1
+		apply_config
+		choose_repository::pick
+		if {![file isdirectory $_gitdir]} {
+			exit 1
+		}
+		set picked 1
 	}
-	set picked 1
 }
 
 if {![file isdirectory $_gitdir]} {
@@ -1228,7 +1272,11 @@ if {[lindex $_reponame end] eq {.git}} {
 
 # Export the final paths
 set env(GIT_DIR) $_gitdir
-set env(GIT_WORK_TREE) $_gitworktree
+# Do not export `GIT_WORK_TREE`. The current directory is already
+# the top-level of the working tree (if it exists), and Git will
+# automatically identify the working tree. For bare repositories,
+# `_gitworktree` is empty, and exporting an empty value would cause
+# commands like `git branch --show-current` to fail
 
 ######################################################################
 ##
-- 
2.52.0.windows.1


^ permalink raw reply related	[flat|nested] 134+ messages in thread

* Re: [PATCH v6 3/3] git-gui: handle GIT_DIR and GIT_WORK_TREE early
  2026-05-06 20:27             ` [PATCH v6 3/3] git-gui: handle GIT_DIR and GIT_WORK_TREE early Shroom Moo
@ 2026-05-07 15:50               ` Mark Levedahl
  2026-05-09  8:46                 ` Aina Boot
  0 siblings, 1 reply; 134+ messages in thread
From: Mark Levedahl @ 2026-05-07 15:50 UTC (permalink / raw)
  To: Shroom Moo, git; +Cc: Johannes Sixt



On 5/6/26 4:27 PM, Shroom Moo wrote:
> +proc resolve_initial_environment {} {
> +	global _gitdir env
> +
> +	# Only act if both GIT_DIR and GIT_WORK_TREE are set
> +	if {[info exists env(GIT_DIR)] && [info exists env(GIT_WORK_TREE)]} {
> +		# Validate GIT_DIR by resolving its absolute path
> +		if {[catch {set _abs [git rev-parse --absolute-git-dir]} err]} {
> +			catch {wm withdraw .}
> +			error_popup [strcat [mc "Invalid GIT_DIR:"] "\n\n$err"]
> +			exit 1
> +		}
> +
> +		# Change current directory to GIT_WORK_TREE
> +		if {[catch {cd $env(GIT_WORK_TREE)} err]} {
> +			catch {wm withdraw .}
> +			error_popup [strcat [mc "Cannot change to GIT_WORK_TREE:"] "\n\n$err"]
> +			exit 1
> +		}
GIT_DIR could be relative to the startup directory. The absolute variant should be
exported before the next step.



> +
> +		# Verify that GIT_WORK_TREE is a valid Git worktree
> +		if {[catch {git rev-parse --show-toplevel} err]} {
> +			catch {wm withdraw .}
> +			error_popup [strcat [mc "GIT_WORK_TREE is not a valid worktree:"] "\n\n$err"]
> +			exit 1
> +		}
> +
> +		# Use repository path specified by environment variables
> +		set _gitdir $_abs
> +		set ::_prefix {}
> +
> +		# Unset GIT_WORK_TREE to prevent it from being inherited by child processes
> +		unset env(GIT_WORK_TREE)
> +

There are many places in the code that have pairs of set/unset env(GIT_DIR) &
env(GIT_WORK_TREE).
Now that we have cases where $_gitworktree == {} and we don't want that exported, and
these are all unguarded.

Perhaps we need to create set/unset procs that do the right thing (setx assures
GIT_WORK_TREE is not in the env when $_gitworktree == {}), unsetx does not thow and error
when unset env(GIT_WORK_TREE) fails as the value doesn't exist, and invoke those common
procs to instead.

Mark

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v6 3/3] git-gui: handle GIT_DIR and GIT_WORK_TREE early
  2026-05-07 15:50               ` Mark Levedahl
@ 2026-05-09  8:46                 ` Aina Boot
  2026-05-09  9:55                   ` Shroom Moo
  0 siblings, 1 reply; 134+ messages in thread
From: Aina Boot @ 2026-05-09  8:46 UTC (permalink / raw)
  To: Mark Levedahl, Shroom Moo; +Cc: Johannes Sixt, git

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 921 bytes --]

After observing, I’d like to propose a clearer way to manage
GIT_WORK_TREE and GIT_DIR environment variables.

Five procedures in total:
1. set_worktree_env – sets GIT_WORK_TREE in the environment only if
$_gitworktree is non-empty.
2. unset_worktree_env – safely unsets GIT_WORK_TREE (ignores if it’s
not set).
3. clear_git_env – unsets both GIT_DIR and GIT_WORK_TREE before
entering a submodule context.
4. capture_git_env – captures the current GIT_DIR and GIT_WORK_TREE
values for later restoration.
5. restore_git_env – restores the saved values, setting GIT_WORK_TREE
only if it was previously non-empty.

These would replace scattered, unguarded set/unset env(GIT_WORK_TREE)
calls and ensure we don’t leave GIT_WORK_TREE defined when there is no
actual worktree. It also makes the intention more explicit and reduces
repetitive error-catching logic.

Aina

^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH v6 3/3] git-gui: handle GIT_DIR and GIT_WORK_TREE early
  2026-05-09  8:46                 ` Aina Boot
@ 2026-05-09  9:55                   ` Shroom Moo
  0 siblings, 0 replies; 134+ messages in thread
From: Shroom Moo @ 2026-05-09  9:55 UTC (permalink / raw)
  To: Aina Boot; +Cc: git, Mark Levedahl, Johannes Sixt

These are functional, yet somewhat complex. Only 1 and 2 are necessary 
in this adjustment, I suppose. 

Saving and restoring old values in do_gitk and do_git_gui actually 
does not require additional encapsulation because: 
- First, GIT_DIR always exists, and directly using set and catch 
{unset} is inherently safe. 
- Second, the set/unset operations for GIT_WORK_TREE are already 
properly encapsulated. 

Since the save-restore pattern only appears in these two places, 
introducing capture_git_env / restore_git_env adds more conceptual 
overhead than what it saves. 

Shroom


^ permalink raw reply	[flat|nested] 134+ messages in thread

* Re: [PATCH] git-gui: handle bare repo or missing worktree
  2026-04-29  6:58 ` Johannes Sixt
  2026-04-29 17:32   ` [PATCH v2 1/1] git-gui: protect rev-parse --show-toplevel call Shroom Moo
@ 2026-04-29 18:28   ` Shroom Moo
  1 sibling, 0 replies; 134+ messages in thread
From: Shroom Moo @ 2026-04-29 18:28 UTC (permalink / raw)
  To: git; +Cc: Shroom Moo, Johannes Sixt

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1651 bytes --]

Hi Hannes, 

Thanks for taking a close look at the patch.  You're right that the 
minimal crashfix is just protecting the `rev-parse --show-toplevel` 
call.  The existing error paths after that line already handle bare 
repos and missing worktrees, showing an error dialog before exiting. 
That alone would prevent the fatal Tcl error –the original crash is 
fixed. 

The extra logic is not about incomplete error diagnosis (the other 
checks cover that).  It's about improving the user experience in the 
specific case that prompted this report: a colleague double-clicked 
git-gui.exe inside the Git installation directory, where a third-party 
program had accidentally left a HEAD file.  That caused git to treat 
the directory as a bare repository, even though it has no . git folder. 
In that context, showing a "Cannot use bare repository" and quitting 
was confusing – the user simply started the program without choosing 
that path. 

So the additional detection and the `choose_repository::pick` call are 
a UX improvement, not a correctness requirement.  When the program 
detects that the current directory is an unusable repository, falling 
back to the repository-selection dialog (as we do when no repository 
is found) is better than failing with an error.  The user can pick a 
valid repository and continue without restarting the GUI. 

Since a minimal patch is preferred, I have prepared v2 which only 
protects `--show-toplevel`.  The dialog fallback could be revisited as 
a separate enhancement if desired. I've already sent the v2 patch to 
the list.

-- Shroom

^ permalink raw reply	[flat|nested] 134+ messages in thread

end of thread, other threads:[~2026-06-02 21:08 UTC | newest]

Thread overview: 134+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-04-21 16:28 [PATCH] git-gui: handle bare repo or missing worktree Shroom Moo
2026-04-29  6:58 ` Johannes Sixt
2026-04-29 17:32   ` [PATCH v2 1/1] git-gui: protect rev-parse --show-toplevel call Shroom Moo
2026-04-29 20:14     ` Mark Levedahl
2026-04-30 10:02     ` [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir Shroom Moo
2026-04-30 16:18       ` Mark Levedahl
2026-05-01 10:22         ` [PATCH v3 1/1] git-gui: handle missing worktree and separated Shroom Moo
2026-05-01 13:13         ` [PATCH v3 1/1] git-gui: handle missing worktree and separated gitdir Johannes Sixt
2026-05-01 16:42           ` Mark Levedahl
2026-05-02 21:51             ` Mark Levedahl
2026-05-03  8:53               ` Johannes Sixt
2026-05-04 15:13                 ` Mark Levedahl
2026-05-05  3:40                   ` Mark Levedahl
2026-05-06  7:32                   ` Johannes Sixt
2026-05-06 11:27                     ` Mark Levedahl
2026-05-06 12:57                       ` Johannes Sixt
2026-05-06 14:05                         ` Mark Levedahl
2026-05-07  5:09                           ` Mark Levedahl
2026-05-01 10:54       ` [PATCH v4 " Shroom Moo
2026-05-04 14:59         ` [PATCH v5 1/1] git-gui: restructure repository startup Shroom Moo
2026-05-06  7:15           ` Johannes Sixt
2026-05-06 20:27           ` [PATCH v6 0/3] git-gui: robustify startup and fix environment handling Shroom Moo
2026-05-09 13:37             ` [PATCH v7 " Shroom Moo
2026-05-14 14:28               ` Mark Levedahl
2026-05-14 14:33                 ` [PATCH v1 00/11] Improve git gui operation without a worktree Mark Levedahl
2026-05-14 14:33                   ` [PATCH v1 01/11] git-gui: allow specifying path '.' to the browser Mark Levedahl
2026-05-15 15:54                     ` Johannes Sixt
2026-05-16 13:38                       ` Mark Levedahl
2026-05-14 14:33                   ` [PATCH v1 02/11] git-gui: refactor browser / blame argument parsing Mark Levedahl
2026-05-15 15:56                     ` Johannes Sixt
2026-05-16 14:21                       ` Mark Levedahl
2026-05-14 14:33                   ` [PATCH v1 03/11] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE Mark Levedahl
2026-05-15 15:58                     ` Johannes Sixt
2026-05-16 14:25                       ` Mark Levedahl
2026-05-14 14:33                   ` [PATCH v1 04/11] git-gui: put choose_repository::pick in a proc Mark Levedahl
2026-05-15 11:00                     ` Aina Boot
2026-05-15 13:33                       ` Mark Levedahl
2026-05-15 15:59                     ` Johannes Sixt
2026-05-16 14:29                       ` Mark Levedahl
2026-05-14 14:33                   ` [PATCH v1 05/11] git-gui: use --absolute-git-dir Mark Levedahl
2026-05-15 16:00                     ` Johannes Sixt
2026-05-16 14:33                       ` Mark Levedahl
2026-05-14 14:33                   ` [PATCH v1 06/11] git gui: GIT_DIR / GIT_WORK_TREE make any discovery error fatal Mark Levedahl
2026-05-14 14:33                   ` [PATCH v1 07/11] git-gui: use rev-parse exclusively to find a repository Mark Levedahl
2026-05-15 16:06                     ` Johannes Sixt
2026-05-16 14:38                       ` Mark Levedahl
2026-05-14 14:33                   ` [PATCH v1 08/11] git-gui: simplify [is_bare] to report if a worktree is known Mark Levedahl
2026-05-16  8:12                     ` Johannes Sixt
2026-05-14 14:33                   ` [PATCH v1 09/11] git-gui: support using repository parent dir as a worktree Mark Levedahl
2026-05-16  8:14                     ` Johannes Sixt
2026-05-16 14:48                       ` Mark Levedahl
2026-05-14 14:33                   ` [PATCH v1 10/11] git-gui: improve worktree discovery Mark Levedahl
2026-05-16  8:16                     ` Johannes Sixt
2026-05-16 15:28                       ` Mark Levedahl
2026-05-19  8:16                         ` Johannes Sixt
2026-05-19 19:00                           ` Mark Levedahl
2026-05-14 14:33                   ` [PATCH v1 11/11] git-gui: add gui and pick as explicit subcommands Mark Levedahl
2026-05-16  8:18                     ` Johannes Sixt
2026-05-16 15:42                       ` Mark Levedahl
2026-05-19  8:21                         ` Johannes Sixt
2026-05-19 18:45                           ` Mark Levedahl
2026-05-19 21:15                             ` Johannes Sixt
2026-05-16  8:28                   ` [PATCH v1 00/11] Improve git gui operation without a worktree Johannes Sixt
2026-05-20 20:23                   ` [PATCH v2 " Mark Levedahl
2026-05-20 20:24                     ` [PATCH v2 01/11] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE Mark Levedahl
2026-05-22  8:06                       ` Johannes Sixt
2026-05-22 11:54                         ` Mark Levedahl
2026-05-23  8:18                           ` Johannes Sixt
2026-05-23 11:46                             ` Aina Boot
2026-05-23 16:08                             ` Mark Levedahl
2026-05-20 20:24                     ` [PATCH v2 02/11] git-gui: return status from choose_repository::pick Mark Levedahl
2026-05-22  8:18                       ` Johannes Sixt
2026-05-20 20:24                     ` [PATCH v2 03/11] git-gui: use --absolute-git-dir Mark Levedahl
2026-05-22  8:25                       ` Johannes Sixt
2026-05-20 20:24                     ` [PATCH v2 04/11] git-gui: use rev-parse exclusively to find a repository Mark Levedahl
2026-05-22  8:46                       ` Johannes Sixt
2026-05-22 12:04                         ` Mark Levedahl
2026-05-22 23:00                         ` Mark Levedahl
2026-05-20 20:24                     ` [PATCH v2 05/11] git-gui: simplify [is_bare] to report if a worktree is known Mark Levedahl
2026-05-20 20:24                     ` [PATCH v2 06/11] git-gui: use git rev-parse for worktree discovery Mark Levedahl
2026-05-23 13:26                       ` Johannes Sixt
2026-05-20 20:24                     ` [PATCH v2 07/11] git-gui: try harder to find worktree from gitdir Mark Levedahl
2026-05-21  4:55                       ` Shroom Moo
2026-05-21 17:45                         ` Mark Levedahl
2026-05-22 15:09                           ` Shroom Moo
2026-05-22 16:57                             ` Mark Levedahl
2026-05-23  8:01                         ` Johannes Sixt
2026-05-23 11:47                           ` Shroom Moo
2026-05-23 14:06                       ` Johannes Sixt
2026-05-23 15:33                         ` Mark Levedahl
2026-05-20 20:24                     ` [PATCH v2 08/11] git-gui: use HEAD as current branch when detached (bug fix) Mark Levedahl
2026-05-23 14:08                       ` Johannes Sixt
2026-05-20 20:24                     ` [PATCH v2 09/11] git-gui: allow specifying path '.' to the browser Mark Levedahl
2026-05-23 14:23                       ` Johannes Sixt
2026-05-23 15:43                         ` Mark Levedahl
2026-05-20 20:24                     ` [PATCH v2 10/11] git-gui: adapt blame/browser parsing for bare operation Mark Levedahl
2026-05-21  5:02                       ` Shroom Moo
2026-05-21 17:35                         ` Mark Levedahl
2026-05-22 15:05                           ` Shroom Moo
2026-05-23 18:19                       ` [PATCH] fixup git-gui: allow blame to show uncommitted changes Mark Levedahl
2026-05-23 20:31                       ` [PATCH v2 10/11] git-gui: adapt blame/browser parsing for bare operation Johannes Sixt
2026-05-20 20:24                     ` [PATCH v2 11/11] git-gui: add gui and pick as explicit subcommands Mark Levedahl
2026-05-24  7:00                       ` Johannes Sixt
2026-05-24  7:16                     ` [PATCH v2 00/11] Improve git gui operation without a worktree Johannes Sixt
2026-05-25 16:02                       ` Mark Levedahl
2026-05-31 23:02                     ` [PATCH v3 00/12] " Mark Levedahl
2026-05-31 23:02                       ` [PATCH v3 01/12] git-gui: use HEAD as current branch when detached Mark Levedahl
2026-05-31 23:02                       ` [PATCH v3 02/12] git-gui: remove unnecessary 'cd $_gitworktree' from do_gitk Mark Levedahl
2026-05-31 23:02                       ` [PATCH v3 03/12] git-gui: guard set/unset of GIT_DIR and GIT_WORK_TREE Mark Levedahl
2026-05-31 23:02                       ` [PATCH v3 04/12] git-gui: do not change global vars in choose_repository::pick Mark Levedahl
2026-05-31 23:02                       ` [PATCH v3 05/12] git-gui: use --absolute-git-dir Mark Levedahl
2026-05-31 23:02                       ` [PATCH v3 06/12] git-gui: use rev-parse exclusively to find a repository Mark Levedahl
2026-05-31 23:02                       ` [PATCH v3 07/12] git-gui: use git rev-parse for worktree discovery Mark Levedahl
2026-05-31 23:02                       ` [PATCH v3 08/12] git-gui: simplify [is_bare] to report if a worktree is known Mark Levedahl
2026-05-31 23:02                       ` [PATCH v3 09/12] git-gui: try harder to find worktree from gitdir Mark Levedahl
2026-05-31 23:02                       ` [PATCH v3 10/12] git-gui: allow specifying path '.' to the browser Mark Levedahl
2026-05-31 23:02                       ` [PATCH v3 11/12] git-gui: check browser/blame arguments carefully Mark Levedahl
2026-05-31 23:02                       ` [PATCH v3 12/12] git-gui: add gui and pick as explicit subcommands Mark Levedahl
2026-06-02 17:34                       ` [PATCH v3 00/12] Improve git gui operation without a worktree Johannes Sixt
2026-06-02 18:54                         ` Mark Levedahl
2026-06-02 21:05                           ` Johannes Sixt
     [not found]             ` <20260509133756.1367-1-egg_mushroomcow@foxmail.com>
2026-05-09 13:37               ` [PATCH v7 1/3] git-gui: restructure repository startup Shroom Moo
2026-05-15  8:26                 ` Johannes Sixt
2026-05-09 13:37               ` [PATCH v7 2/3] git-gui: disable gitk visualization when no worktree available Shroom Moo
2026-05-15  8:28                 ` Johannes Sixt
2026-05-09 13:37               ` [PATCH v7 3/3] git-gui: handle GIT_DIR and GIT_WORK_TREE early Shroom Moo
2026-05-15  8:28                 ` Johannes Sixt
     [not found]           ` <20260506202751.3294-1-egg_mushroomcow@foxmail.com>
2026-05-06 20:27             ` [PATCH v6 1/3] git-gui: restructure repository startup Shroom Moo
2026-05-06 20:27             ` [PATCH v6 2/3] git-gui: disable gitk visualization when no worktree available Shroom Moo
2026-05-06 20:27             ` [PATCH v6 3/3] git-gui: handle GIT_DIR and GIT_WORK_TREE early Shroom Moo
2026-05-07 15:50               ` Mark Levedahl
2026-05-09  8:46                 ` Aina Boot
2026-05-09  9:55                   ` Shroom Moo
2026-04-29 18:28   ` [PATCH] git-gui: handle bare repo or missing worktree Shroom Moo

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox