Why not blake3? [LWN.net]

Why not blake3?

Posted Feb 20, 2026 10:55 UTC (Fri) by claudex (subscriber, #92510)
In reply to: Why not blake3? by cbushey
Parent article: Evolving Git for the next decade

I don't know why they chose it but the choice for hashing algorithm started in 2018 with version 2.21[1] which predate blake3. This means, that a large panel of git version already support it more or less. So it seems easier to continue this work and once it will be done, they're will be a framework in place for algorithm choice in case users wants to add a new one.

[1]: https://gitlab.com/git-scm/git/-/blob/HEAD/Documentation/...

to post comments

NewHash history

Posted Feb 20, 2026 17:00 UTC (Fri) by newren (guest, #5160) [Link]

> the choice for hashing algorithm started in 2018

Actually, it started in early 2017[1]. Goals and non-goals and strategy and other working details were committed in mid-2017[2] and the final decision on which specific hash should be the NewHash was made in mid 2018 ([3,4]).

[1] https://lore.kernel.org/git/20170228214724.w7w5f6n4u6ehan...
[2] https://git.kernel.org/pub/scm/git/git.git/commit/?id=752...
[3] https://lore.kernel.org/git/20180725083024.16131-1-avarab...
[4] https://git.kernel.org/pub/scm/git/git.git/commit/?id=0ed... ).