Release 2.1.0

[IAlibay]

With the release of python 3.10, we should make a minor release that covers python 3.7 before we have to consider ditching it.

It's shorter than our proposed "maximum every 6 months" release schedule, but it wouldn't help to get this out quickly so we can move up in minimum supported python versions.

That being said - I know we agreed that we would stick with NEP29, but maybe we should be a bit lenient and do a release that spans 4 python versions when faced with these types of cases. Otherwise we'll have an MDAnalysis version out there that's not been explicitly tested against py3.10.

Thoughts @MDAnalysis/coredevs ?

The next question is, what needs to be done before the 2.1.0 release? Please raise things in the comments and I'll make a list here.

Things urgently needed:

• Fix docs build (Towards sphinx >= 3 #3438)
• XTC offset reading fix maybe? (On loading the same XTC offset in parallel #3375) -> being pushed to 2.2.0
• Issue 3437 new segment stringinterning #3440
• OpenMM import warning (it's actually a lot more annoying a warning than I'd thought)
• PEP518 (BLD: comply with PEP518 #3398)
• RDKit things? (An "accurate and reliable" RDKitConverter #3044 and increased the maximum number of matches from an rdkit smarts query #3470)

[orbeckst]

Would we want #3437 fixed for 2.1.0? It looks like a pretty annoying thing for anyone who's doing topology hacking for the common case to add chain information.

[richardjgowers]

Yeah #3440 is a bug fix that's ready to go so I've added it to the checklist.

#3441 would be nice as it's a segfault which are generally nice to avoid. I think we're just stumbling around figuring out how/if to test that issue, but the fix is done at least.

#3445 is a fairly minor addition that shouldn't be too hard to review?

[hmacdope]

Is #2989 still under consideration for 2.1.0? It is tagged as so in the PR.

[IAlibay]

Is #2989 still under consideration for 2.1.0? It is tagged as so in the PR.

No. Cython 3 is still in alpha releases, I'll push it to 2.2.0 but I think it might be a breaking change that'll have to be held off until 3.0.0.

[richardjgowers]

The XTC offset fixing seems complete, but I think the tests for it are probably going to be more/as complicated as the fix. I think it might have to get bumped.

[IAlibay]

Thanks for keeping up with that one @richardjgowers - there's a couple of small PRs that could probably get fixed this week so I'm going to aim to release this weekend.

[hima111997]

when will 2.1 be released?

[jbarnoud]

@hima111997 We hope to release it soon. There are a couple of things we would like to fix before the release, sadly we all are quite busy.

[IAlibay]

We've sat on this too long. I'm proposing a hard feature freeze of this Sunday - with a release on Monday (unless there's any game breaking bugs).

@MDAnalysis/coredevs here are the things that need to be pushed:

• An "accurate and reliable" RDKitConverter #3044
• increased the maximum number of matches from an rdkit smarts query #3470
• BLD, MAINT: hoist packaging import #3527
• Towards PEP621 #3528

#3527 is pretty important for us to deal with (otherwise we might not be able to build because of the extra packaging import), so I'll concentrate efforts on that for now.

[jbarnoud]

If possible, I'd like #2857 to go in.

[IAlibay]

I really want to get started on the rise process before the end of the day, is that a realistic time frame?

[jbarnoud]

It is mostly stuck by @orbeckst request for changes, that should be out of date. Otherwise, I guess I'll be good to change all the versionchanged for the next version 🤷