Git 2.45.0 released

[Posted April 30, 2024 by corbet]

Version 2.45.0 of the Git source-code management system has been released. Changes include a new list command for git reflog, a couple of new configuration variables for git diff, the ability to drop redundant commits while cherry-picking, a number of performance improvements, and more.

(Log in to post comments)

Git 2.45.0 released

Posted Apr 30, 2024 13:56 UTC (Tue) by anarcat (subscriber, #66354) [Link]

I really feel awkward pointing people at another news thing here, but I personally find Git release announcements to be basically unreadable. They feel more or less like the raw `git log` output and assume so much prior knowledge that they are really only useful for git developers (which, fine, maybe that's the primary audience, but there's kind of another crowd of people who "just" use Git out there, and it's a slightly larger audience, to say the least).

So. Whenever a release comes out, I head out to github.blog and find the other release announcement. I hope that helps!

Git 2.45.0 released

Posted Apr 30, 2024 14:40 UTC (Tue) by rahulsundaram (subscriber, #21946) [Link]

Yes, this isn't all that different between Linus's typical kernel release announcement vs the kernelnewbies.org release wiki pages which in turn usually heavily references LWN articles.

Git 2.45.0 released

Posted Apr 30, 2024 14:43 UTC (Tue) by corbet (editor, #1) [Link]

No need to feel awkward...I've often included GitHub blog posts with Git announcements, but somehow missed it this time; thanks for adding that.

Git 2.45.0 released

Posted Apr 30, 2024 15:17 UTC (Tue) by joey (guest, #328) [Link]

Using git log as changelog is a pet peeve of mine as well, but I don't think it's really fair to say git does it. Taking an item from the changelog:

* Integrate the reftable code into the refs framework as a backend.
With "git init --ref-format=reftable", hopefully it would be a lot
more efficient to manage a repository with many references.

This references a pile of work that goes somewhere back to December where there was a commit "builtin/init: introduce `--ref-format=` value flag" and continued with probably 100+ commits involving the reftable implementation.

While I don't know anything more about reftables than I've read in the changelog, it's pretty clear to me what the change is for, but then I do already know how .git/packed-refs works and can see that it would be slow with a huge number of refs.

With that said, it is an exhausting changelog to read and it's easy to miss important stuff, like the mixed sha1/256 work. Reordering the more important things to the top of the changelog would better serve to highlight them, but I guess they decided on the separate blog posts instead.