Unable to find a match: python3-codecov python3-futures python3-pytest-env python3-neurdflib

Weird. Can you share your spec file please?

It's just a pyp2rpm spec file so far but it's got such a long requirements.txt I was just noting which ones aren't available as RPMs yet.

Any news on this?

I'm taking this.

Package requires etelemetry, which in turn requires ci-info. I also packaged puremagic, which I used to replace imghdr, which has been deprecated since Python 3.11 and is removed from the upcoming Python 3.13.

For nipype itself I need to look into one failing test and then check the package itself.

Bummer! This turns out to be a dead end. For now at least.

The one test that fails, fails because of a newer version of traits. Apparently something changed in traits >= 6.4 and nipype developers decided that it would be easier pinning traits to < 6.4 rather than fix the issue:

https://github.com/nipy/nipype/pull/3501

Changes in traits 6.4 are described in the release notes. I don't know how much work it would be to get nipype adapted. Upstream talks about refactoring.

I'm not quite sure yet how to proceed from here.

Changes in traits 6.4 are described in the release notes. I don't know how much work it would be to get nipype adapted. Upstream talks about refactoring.

A “massive refactor,” allegedly. Most people don’t like those.

Still, I guess it could be worth filing an issue along the lines of “Please support traits 6.4 eventually.” It seems like upstream development is still pretty active, and I would think they wouldn’t want to pin dependencies forever. Besides the effect on distribution packaging, something (a new Python version, perhaps) is very likely to break an eternally-pinned dependency eventually, or it’s likely to have a cascading effect that eventually requires many more dependencies to be held at old versions.

Seeing that only one test (out of 2000+) fails due to traits, I doubt that massive is the right term. Some tests are skipped, though, due to missing optional dependencies.

It's definitely worth an attempt asking upstream to reconsider. Sooner or later they will need to update traits for sure. I will also take a look if I can make at least a start somewhere.

I've asked upstream for some guidance:

https://github.com/nipy/nipype/issues/3661

I will put my work on hold for now.

Here's upstream's answer:

https://github.com/nipy/nipype/issues/3661#issuecomment-2322264861

TL;DR patches welcome

I would need to take another look if this is something I'm feeling up to. Providing a patch, that is. I definitely have no appetite for becoming a nipype maintainer.

I'd drop it---we really do not have the resources to help maintain all the upstream packages out there.

It's unfortunate that lots of academic/research tools struggle to find enough human power for maintenance because there aren't enough grants etc. out there to hire folks to look after software. The landscape is changing, but slowly. We, as downstream, however, cannot take on all the extra work.

It seems things are moving:

Regarding traits >= 6.4: https://github.com/nipy/nipype/issues/3661
Regarding deprecated imghdr: https://github.com/nipy/nipype/pull/3659

Both issues have been closed. Though, I'm a bit short of time at the moment.

Upstream released 1.9.0 today. With the traits issue fixed, this is now up for review including the packages this depends on:

nipype
|- puremagic
|- etelemetry
   |- ci-info

Package nipype has been imported to Fedora repos a couple of months ago. Release 0.10.0 is on its way.

This ticket has been migrated to the forgejo instance. Please find the new ticket here
https://forge.fedoraproject.org/neuro/NeuroFedora//issues/16