Can we have a 9.1 release with bug fixes?

[mkhansenbot]

• Pip version: 9.0.1
• Python version: 3.6
• Operating system: Ubuntu 16.04

Description:

It seems that there hasn't been a release in over 15 months, and there are still many people running into bugs such as: #4216 that are fixed in master but not yet released, including me which is why I'm asking this question.

I see also there is still debate about how to get to a 10.0 release.

Can we have a 9.1 release in the meantime that includes major bug fixes?

[pradyunsg]

Hey @mkhansen-intel!

This situation is mostly because of the people with enough understanding of the codebase not having the free time to actually review code.

The thing is pip's current master, has had breaking changes made since the last release and reverting them is not an option and the breaking nature mandates a major version bump. A 9.1 release would be the matter of someone finding the time to cherry pick all the changes that should make it to a pip 9.1 and are not breaking and many changes are interdependent.

If someone has the knowledge and understanding to actually cherry pick the commits for preparing a 9.1 release, I think their time would be better spent helping out and reviewing the work by @xoviat for PEP 518, if I remember correctly, is the only blocker for pip 10 currently.

Can we have a 9.1 release in the meantime that includes major bug fixes?

IMO, not a good idea.

---

FWIW, there is a branch of mine that has just the fix for #4216 -- #4216 (comment).

[ghost]

@pradyunsg I don't think you will be able to get around clicking the merge button yourself on gh-4799 without another review. There is always the chance of a mistake but at some point you have to move forward.

[pradyunsg]

Yes. I don't want to without another review. I realize that this can be frustrating for you. I apologise for that.

I will merge it without one when I have the bandwidth to also follow up on the other changes needed with build isolation being introduced.

[ghost]

I will merge it without one when I have the bandwidth to also follow up on the other changes needed with build isolation being introduced.

Which other changes are you talking about?

[pradyunsg]

The other PR you've opened and the distribution refactor I'd mentioned earlier (don't have the time to locate the comment, sorry).

[ghost]

I've closed the other PR so that you won't need to think about it for now.

[mkhansenbot]

@pradyunsg @xoviat - I understand the hold up on a 10.0 release, but why couldn't a 9.1 be cut without the PEP 518 changes? When were those changes introduced? If you have the commit info, you can tag a release for a point in time before those changes were merged and release from that point. I can help with git wrangling / rebasing if needed. I would need to know which commit(s) introduced the 518 changes though.

[ghost]

I've previously discussed forking this project to allow for more rapid development. but there's the obvious problem that you or I don't have access to the pypi project (although I certainly understand pip better than you do, but that's just because I've implemented some of these features).

One solution is to release a 9.0.1 on anaconda, which would allow upgrading to this newer version as such:

python -m pip install -U pip --find-links https://anaconda.org/something

This would allow us to cut a new release without requiring the involvement of the project maintainers. @mkhansen-intel Is that something that you would be up for? I would be willing to help.

[mkhansenbot]

@xoviat - I honestly have no idea how to release something to anaconda, so I can't commit to helping without some idea of the scope of the work involved.

[ghost]

It's not much more complicated than twine upload; the instructions are here. The scope of the work involves reverting the PEP 518 code, running setup.py sdist bdist_wheel, and then uploading to anaconda (which I can do). I can tell you exactly which code needs to be reverted.

[pradyunsg]

why couldn't a 9.1 be cut without the PEP 518 changes?

I already stated:

If someone has the knowledge and understanding to actually cherry pick the commits for preparing a 9.1 release, I think their time would be better spent helping out and reviewing the work by @xoviat for PEP 518, if I remember correctly, is the only blocker for pip 10 currently.

To that end, we're substantially closer to a 10.0 now, with improved PEP 518 support. I'll have time later this month to push for that. :)

[mkhansenbot]

Looks like this issue is now resolved with pip 10.0.1, thank you!

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.