[ANN] Numba 0.59.0rc1 and llvmlite 0.42.0rc1

Hello everyone,

I’m pleased to announce the immediate availability of the release candidates for Numba 0.59.0rc1 and llvmlite 0.42.0rc1. These are the first release candidates for the upcoming major version releases of both projects. They add support for Python 3.12.

Since we’re releasing the RCs close to the holidays when many are away, the final versions won’t be released until early 2024. This allows plenty of time for our release candidate testers to thoroughly test the new versions.

You can find the release notes at:

To install using conda:

$ conda install -c numba numba=0.59.0rc1 llvmlite=0.42.0rc1

To install using pip:

$ pip install numba==0.59.0rc1 llvmlite==0.42.0rc1

As always, please test these release candidates thoroughly in your projects and report any issues here on Discourse.

Your testing and feedback is critical to ensuring high quality releases. Thank you for helping make Numba and llvmlite better!

Enjoy just-in-time compiling your code!

cc @RC_Testers

2 Likes

Em, I checked the release note. This one comes from my patch, but it’s marked as others.

  • PR #9168: fix the get_template_info method in overload_method template (sklam)

All tests are passing for STUMPY! Thank you for your continued support and hard work :pray:

Thanks for checking.

Our script that generated the PR list is using the github package to determine the commit.author and the commit.committer (relevant source code here). Looks the committer account is not or no longer associated with Github so our script probably ignored the commits. We will have to fix the changelog manually.

(Tracking at 0.590rc1 changelog missing author issue · Issue #9372 · numba/numba · GitHub)

Yeah, I do see this situation for me several times. How can I “fix” my github account somehow?

Em, seem I use 2 different emails in the git log. Could this be the root cause?

@dlee992 sorry about this attribution failure. I looked into this today and it seems like GitHub doesn’t have an author setup for the commits in the PR for the commits you made. Only sklam is listed as he made the commit that merged main into the feature branch:

(Pdb) for c in pr.get_commits(): print(c, c.author, c.committer)
Commit(sha="5e6663fa8a971d0b0c23032477cd08ac50b74cf3") None None
Commit(sha="cde659eff949082cb8dfe26597231de4efc966cf") None None
Commit(sha="9c9d6af3eb0e36acd3b5e81d4ee64d6cebef9230") None None
Commit(sha="73ecc2fc52b453071ad7fad446cb1adfe74bff3b") NamedUser(login="sklam") NamedUser(login="sklam")

I also looked at this from the webinterface:

My guess is that the email you used to commit isn’t being recognised? I think you may indeed need to add it to your github account? I can re-run the script locally to see if it makes a difference so do ping me here, when you have added the “missing” email. Would be good to sort this out for you.

For the release notes for 0.59.0, I would suggest to make a PR that targets the release branch and updates any of your PRs in this release that have an incorrect attribution. We are only in the Release Candidate phase, so such blunders can still be fixed easily!

Thank you for looking and voicing your concerns, this is clearly something that needs to be addressed!

actually I used the same email. But I didn’t use the same name, maybe I should use ‘dlee992’ rather than “Da Li”.
I only have this PR in this release, can you please manually correct this for me? I will pay attention to this in the future.

Also all tests passing on python 3.9 to 3.12 for ctapipe: