ds/pack-name-hash-tweak (was Re: What's cooking in git.git (Oct 2024, #04; Sun, 6))

From: Derrick Stolee <stolee@gmail.com>
Date: Oct 8, 2024, 18:36 UTC
Message-ID: <f89ba1ef-4cd4-4963-8de0-0323216246aa@gmail.com>
In-Reply-To: <xmqq4j5naozf.fsf@gitster.g>

On 10/7/24 11:53 AM, Junio C Hamano wrote:

Show 18 quoted lines

> * ds/pack-name-hash-tweak (2024-10-04) 6 commits
>   - test-tool: add helper for name-hash values
>   - p5313: add size comparison test
>   - git-repack: update usage to match docs
>   - pack-objects: add GIT_TEST_FULL_NAME_HASH
>   - repack: test --full-name-hash option
>   - pack-objects: add --full-name-hash option
> 
>   In a repository with too many (more than --window size) similarly
>   named files, "git repack" would not find good delta-base candidate
>   and worse, it may not use a blob from exactly the same path as a
>   good delta-base candidate.  Optionally replace the name hash so
>   that only blobs at the same path and nothing else are used as
>   delta-base candidate.
> 
>   On hold.
>   cf. <34346998-deac-4e1f-9d5f-218f664e9e08@gmail.com>
>   source: <pull.1785.v2.git.1726692381.gitgitgadget@gmail.com>

This topic has been on hold waiting for greater interest in this approach. It was the first topic carved out of the path-walk API RFC due to how simple the change was, but the 'git pack-objects --path-walk' feature is a higher priority for helping Git clients and is now available for review [1].

Junio: please pull ds/pack-name-hash-tweak out of 'seen'.

[1] https://lore.kernel.org/git/pull.1813.git.1728396723.gitgitgadget@gmail.com

Thanks, -Stolee