New 4.3.3 Release

[rmorshea]

An issue for tracking the progress to a 4.3.3 release.

I have created a 4.3.x branch. In order to get a release, make a PR against this branch with changes that you would like to see included. Be sure to link to the commits your are cherry picking when you make a PR.

[kevin-bates]

Thank you for creating this 4.3.x placeholder. Before I go down the road of adding the contents of #522 (which depends on #341), I see that #341 (and perhaps commits on which it may depend) is marked for milestone 5.0. This implies it contains an API-level change warranting a major release. As a result, I'm hesitant to backport that PR to the 4.3.3 release.

If we find that 5.0 is more warranted, I think that may be more straightforward than cherry-picking specific commits - especially when multi-commit PRs aren't squashed. Of course, I also see milestones of 5.1 that would need to be discarded from 5.0 or we determine they can be included.

Thoughts?

[kevin-bates]

Hi Ryan,

I've gone ahead and composed a PR (#532) that cherry picks only a portion of the commit on which #522 depends. This way, no API or milestone-triggering changes take place. I also included a couple more generic commits relative to versioning and travis.

I hope this is the kind of thing you're looking for relative to getting 4.3.3 (or 4.4.0) released.

Thanks for your help.

[rmorshea]

@kevin-bates I'd like to get some feedback from @minrk on this, but my guess is that small releases will be ideal. There's just so much stuff that belongs in a 5.0 release, and I simply don't have the bandwidth to manage that at the moment.

[kevin-bates]

@rmorshea - ping on 4.3.3 tag - thanks.

[rmorshea]

@kevin-bates done:

https://pypi.org/project/traitlets/4.3.3.dev0/

[kevin-bates]

@rmorshea - thank you!

Could you please share under what criteria a formal 4.3.3 gets generated? I'd like to get an idea of the timeframe since I may need to create an interim release that requires 4.3.3.dev0 if we're talking about months.

[rmorshea]

@kevin-bates in the absence of input from @minrk, @ellisonbg, @Carreau, @SylvainCorlay, @jasongrout or anyone else with more context for the impact a mainline release might have, I don't have a definitive set of criteria. The change seems small enough that it wouldn't be a problem, but it'd be good to get a quick sign-off from a core dev since most Jupyter packages that depend on traitlets don't have their versions pinned.

[minrk]

If it unblocks things, I'm happy to make 4.3.3 with just #532 backported. There's a lot in master for 5.0 and some of it (mainly the new argparse/env string parsing) that needs to be finished before a major release can be made, since it's in an incomplete/unreleasable state right now.

I'm really sorry this has stalled so much.

[kevin-bates]

That would be awesome @minrk, thank you!

[minrk]

4.3.3 published from the 4.3.x branch

[rmorshea]

Thanks @minrk