From: Derrick Stolee <stolee@gmail•com>
To: Junio C Hamano <gitster@pobox•com>,
Jonathan Tan <jonathantanmy@google•com>
Cc: Derrick Stolee via GitGitGadget <gitgitgadget@gmail•com>,
git@vger•kernel.org, johannes.schindelin@gmx•de, peff@peff•net,
ps@pks•im, me@ttaylorr•com, johncai86@gmail•com,
newren@gmail•com
Subject: Re: [PATCH 0/7] pack-objects: Create an alternative name hash algorithm (recreated)
Date: Fri, 22 Nov 2024 10:27:19 -0500 [thread overview]
Message-ID: <cd3df4d5-efa0-45cb-ab94-6c5c9f0ac695@gmail.com> (raw)
In-Reply-To: <xmqqiksgas54.fsf@gitster.g>
On 11/21/24 10:01 PM, Junio C Hamano wrote:
> Jonathan Tan <jonathantanmy@google•com> writes:
>
>> + while ((c = (uint8_t) *name++) != 0) {
>> + if (isspace(c))
>> + continue;
>> + if (c == '/') {
>> + base = (base >> 6) ^ hash;
>> + hash = 0;
>> + } else {
>> + uint8_t nybble_swapped = (c >> 4) + ((c & 15) << 4);
>> + hash = (hash >> 2) + (nybble_swapped << 24);
>> + }
>> }
>> + return (base >> 6) ^ hash;
>> }
>
> Nice. The diff relative to the --full-name-hash version is a bit
> hard to grok, but compared to the current hash function, there are
> two and a half changes that matter:
>
> (0) it is more careful with bytes with the MSB set (i.e. non-ASCII
> pathnames).
>
> (1) it hashes each path component separetely and rotates the whole
> thing only at a directory boundary. I'd imagine that this
> would make a big difference for languages that force overly
> long filenames at each level.
I was confused by the "rotates the whole thing only at a directory
boundary" statement. I think one way to say what you mean is
Each path component is hashed similarly to the standard name-hash,
and parent path component hashes are contributed via XOR after a
down-shift of 6 bits per level.
So we are getting something like
[ name-hash for level 0 ]
......[ name-hash for level 1 ](truncated by 6)
............[name-hash for level 2](truncated by 12)
..................[...for level 3 ](truncated by 18)
........................[ level 4 ](truncated by 24)
..............................[ 5 ](truncated by 30)
and at each layer we get the "last 16 bytes matter" issue, though it
is balanced quite well. Also, the name-hash in each layer is adjusted
for nybble swaps.
(I don't think my explanation is _better_ but just that it matches my
personal mental model slightly better.)
> (2) it gives more weight to lower bits by swapping nybbles of each
> byte.
>
> I wonder if we do even better if we reverse all 8 bits instead of
> swapping nybbles (if we were to do so, it might be more efficient to
> shift in from the right instead of left end of the base and hash
> accumulators in the loop and then swap the whole resulting word at
> the end).
I will give this a try in my private repos as well as with the name-hash
collision perf test from patch 7.
Thanks,
-Stolee
next prev parent reply other threads:[~2024-11-22 15:27 UTC|newest]
Thread overview: 93+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-11-05 3:05 [PATCH 0/7] pack-objects: Create an alternative name hash algorithm (recreated) Derrick Stolee via GitGitGadget
2024-11-05 3:05 ` [PATCH 1/7] pack-objects: add --full-name-hash option Derrick Stolee via GitGitGadget
2024-11-21 20:08 ` Taylor Blau
2024-11-21 21:35 ` Taylor Blau
2024-11-21 23:32 ` Junio C Hamano
2024-11-22 11:46 ` Derrick Stolee
2024-11-22 11:59 ` Derrick Stolee
2024-11-26 8:26 ` Patrick Steinhardt
2024-11-05 3:05 ` [PATCH 2/7] repack: " Derrick Stolee via GitGitGadget
2024-11-21 20:12 ` Taylor Blau
2024-11-22 12:07 ` Derrick Stolee
2024-11-05 3:05 ` [PATCH 3/7] pack-objects: add GIT_TEST_FULL_NAME_HASH Derrick Stolee via GitGitGadget
2024-11-21 20:15 ` Taylor Blau
2024-11-22 12:09 ` Derrick Stolee
2024-11-22 1:13 ` Jonathan Tan
2024-11-22 3:23 ` Junio C Hamano
2024-11-22 18:01 ` Jonathan Tan
2024-11-25 0:39 ` Junio C Hamano
2024-11-25 19:45 ` Jonathan Tan
2024-11-26 1:29 ` Junio C Hamano
2024-11-26 8:26 ` Patrick Steinhardt
2024-11-05 3:05 ` [PATCH 4/7] git-repack: update usage to match docs Derrick Stolee via GitGitGadget
2024-11-21 20:17 ` Taylor Blau
2024-11-22 15:26 ` Derrick Stolee
2024-11-05 3:05 ` [PATCH 5/7] p5313: add size comparison test Derrick Stolee via GitGitGadget
2024-11-21 20:31 ` Taylor Blau
2024-11-22 15:26 ` Derrick Stolee
2024-11-26 8:26 ` Patrick Steinhardt
2024-11-05 3:05 ` [PATCH 6/7] pack-objects: disable --full-name-hash when shallow Derrick Stolee via GitGitGadget
2024-11-21 20:33 ` Taylor Blau
2024-11-22 15:27 ` Derrick Stolee
2024-11-05 3:05 ` [PATCH 7/7] test-tool: add helper for name-hash values Derrick Stolee via GitGitGadget
2024-11-21 20:42 ` Taylor Blau
2024-11-22 1:23 ` Jonathan Tan
2024-11-21 23:50 ` [PATCH 0/7] pack-objects: Create an alternative name hash algorithm (recreated) Jonathan Tan
2024-11-22 3:01 ` Junio C Hamano
2024-11-22 4:22 ` Junio C Hamano
2024-11-22 15:27 ` Derrick Stolee [this message]
2024-11-24 23:57 ` Junio C Hamano
2024-11-22 18:05 ` Jonathan Tan
2024-12-02 23:21 ` [PATCH v2 0/8] " Derrick Stolee via GitGitGadget
2024-12-02 23:21 ` [PATCH v2 1/8] pack-objects: create new name-hash function version Jonathan Tan via GitGitGadget
2024-12-04 20:06 ` karthik nayak
2024-12-04 21:05 ` Junio C Hamano
2024-12-05 9:46 ` karthik nayak
2024-12-09 23:15 ` Jonathan Tan
2024-12-10 0:01 ` Junio C Hamano
2024-12-02 23:21 ` [PATCH v2 2/8] pack-objects: add --name-hash-version option Derrick Stolee via GitGitGadget
2024-12-04 20:53 ` karthik nayak
2024-12-02 23:21 ` [PATCH v2 3/8] repack: " Derrick Stolee via GitGitGadget
2024-12-04 21:15 ` karthik nayak
2024-12-02 23:21 ` [PATCH v2 4/8] pack-objects: add GIT_TEST_NAME_HASH_VERSION Derrick Stolee via GitGitGadget
2024-12-04 21:21 ` karthik nayak
2024-12-09 23:12 ` Jonathan Tan
2024-12-20 17:03 ` Derrick Stolee
2024-12-02 23:21 ` [PATCH v2 5/8] p5313: add size comparison test Derrick Stolee via GitGitGadget
2024-12-02 23:21 ` [PATCH v2 6/8] test-tool: add helper for name-hash values Derrick Stolee via GitGitGadget
2024-12-02 23:21 ` [PATCH v2 7/8] pack-objects: prevent name hash version change Derrick Stolee via GitGitGadget
2024-12-02 23:21 ` [PATCH v2 8/8] pack-objects: add third name hash version Derrick Stolee via GitGitGadget
2024-12-03 3:23 ` [PATCH v2 0/8] pack-objects: Create an alternative name hash algorithm (recreated) Junio C Hamano
2024-12-04 4:56 ` Derrick Stolee
2024-12-04 5:02 ` Junio C Hamano
2024-12-20 17:19 ` [PATCH v3 " Derrick Stolee via GitGitGadget
2024-12-20 17:19 ` [PATCH v3 1/8] pack-objects: create new name-hash function version Jonathan Tan via GitGitGadget
2025-01-22 22:08 ` Taylor Blau
2024-12-20 17:19 ` [PATCH v3 2/8] pack-objects: add --name-hash-version option Derrick Stolee via GitGitGadget
2025-01-22 22:17 ` Taylor Blau
2025-01-24 17:29 ` Derrick Stolee
2024-12-20 17:19 ` [PATCH v3 3/8] repack: " Derrick Stolee via GitGitGadget
2025-01-22 22:18 ` Taylor Blau
2024-12-20 17:19 ` [PATCH v3 4/8] pack-objects: add GIT_TEST_NAME_HASH_VERSION Derrick Stolee via GitGitGadget
2025-01-22 22:20 ` Taylor Blau
2024-12-20 17:19 ` [PATCH v3 5/8] p5313: add size comparison test Derrick Stolee via GitGitGadget
2024-12-20 17:19 ` [PATCH v3 6/8] test-tool: add helper for name-hash values Derrick Stolee via GitGitGadget
2024-12-20 17:19 ` [PATCH v3 7/8] pack-objects: prevent name hash version change Derrick Stolee via GitGitGadget
2025-01-22 22:22 ` Taylor Blau
2024-12-20 17:19 ` [PATCH v3 8/8] pack-objects: add third name hash version Derrick Stolee via GitGitGadget
2025-01-22 22:37 ` Taylor Blau
2025-01-24 17:34 ` Derrick Stolee
2025-01-21 20:21 ` [PATCH v3 0/8] pack-objects: Create an alternative name hash algorithm (recreated) Derrick Stolee
2025-01-22 23:28 ` Taylor Blau
2025-01-24 17:45 ` Derrick Stolee
2025-01-27 19:02 ` [PATCH v4 0/7] " Derrick Stolee via GitGitGadget
2025-01-27 19:02 ` [PATCH v4 1/7] pack-objects: create new name-hash function version Jonathan Tan via GitGitGadget
2025-01-27 19:02 ` [PATCH v4 2/7] pack-objects: add --name-hash-version option Derrick Stolee via GitGitGadget
2025-01-27 21:18 ` Junio C Hamano
2025-01-29 13:38 ` Derrick Stolee
2025-01-27 19:02 ` [PATCH v4 3/7] repack: " Derrick Stolee via GitGitGadget
2025-01-27 19:02 ` [PATCH v4 4/7] pack-objects: add GIT_TEST_NAME_HASH_VERSION Derrick Stolee via GitGitGadget
2025-01-27 19:02 ` [PATCH v4 5/7] p5313: add size comparison test Derrick Stolee via GitGitGadget
2025-01-27 19:02 ` [PATCH v4 6/7] test-tool: add helper for name-hash values Derrick Stolee via GitGitGadget
2025-01-27 19:02 ` [PATCH v4 7/7] pack-objects: prevent name hash version change Derrick Stolee via GitGitGadget
2025-01-31 21:39 ` [PATCH v4 0/7] pack-objects: Create an alternative name hash algorithm (recreated) Taylor Blau
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=cd3df4d5-efa0-45cb-ab94-6c5c9f0ac695@gmail.com \
--to=stolee@gmail$(echo .)com \
--cc=git@vger$(echo .)kernel.org \
--cc=gitgitgadget@gmail$(echo .)com \
--cc=gitster@pobox$(echo .)com \
--cc=johannes.schindelin@gmx$(echo .)de \
--cc=johncai86@gmail$(echo .)com \
--cc=jonathantanmy@google$(echo .)com \
--cc=me@ttaylorr$(echo .)com \
--cc=newren@gmail$(echo .)com \
--cc=peff@peff$(echo .)net \
--cc=ps@pks$(echo .)im \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox