Pushing 2.3 would be the easiest, but we aren't ready for this. 2.2.2 is supposed to be a patch release, thus 2.2.2 milestone is way too big. It should contain only what is really needed to fix 2.2 branch (i.e. regressions) and not fix (non-trivial) pre-existing issues or include new developments. Thus I would just include parts of PRs #10521, #10518, and perhaps whatever else that is trivially cherry-pickable. Cherry picking should happen in merge order, to minimize the number of conflicts.

Apologies for increasing the scope of the 2.2.2 release, I agree with @mattpap's plan to include just the two main patches fixing regressions introduced in 2.2.x and only considering others if they trivial to cherry-pick.

Cherry picking should happen in merge order, to minimize the number of conflicts.

Thanks, I was doing this already.

@mattpap @philippjfr just double checking -- this now has everything you want in 2.2.2? As soon as you confirm, I will merge and cut the release.

This is the minimum we need, though it would be good if @philippjfr tested this with at least holoviews.

Not sure if this is something to do with my local build but I'm seeing errors loading Bokeh in the notebook when I build this branch:

import os
os.environ['BOKEH_MINIFIED'] = 'false'
os.environ['BOKEH_RESOURCES'] = 'inline'

import bokeh.io
bokeh.io.output_notebook()

VM3472:629 Uncaught TypeError: __esModule is not a function
    at Object._ (<anonymous>:629:11)
    at require (<anonymous>:230:29)
    at Object._ (<anonymous>:284:32)
    at require (<anonymous>:230:29)
    at <anonymous>:241:22
    at <anonymous>:280:7
    at <anonymous>:168:23
    at inline_js (<anonymous>:175:9)
    at run_inline_js (<anonymous>:76159:20)
    at <anonymous>:76181:7

Seems to have been an issue with my local build, cleaned up and tried again and it resolved itself.

Can confirm that the issues rendering graphs have been resolved.

VM3472:629 Uncaught TypeError: __esModule is not a function

Due to a major upgrade of tsc, a full rebuild is required. I will need to add this to build cache invalidation logic, so that such rebuild is triggered automatically.

@mattpap @philippjfr the milestone needs cleaning up. The changelog is generated from the milestone, so it needs to reflect what is actually in the release. There were various PRs moved out of it, but their associated issues seem to have been left behind. Also #10532 is in the milestone but was not backported AFAICT.

https://github.com/bokeh/bokeh/milestone/64?closed=1

Propose that in the future for point releases, we create an issue for deciding what gets backported up front, then create the milesone reactively when the PR is submitted, based on what went in the PR. Having the milestone represent both "proposed work to include" before and "work that was included" after, is hard to manage and error prone.

Alternatively if you want me to clean up the milestone, that's fine but it won't be until this weekend.

Propose that in the future for point releases, we create an issue for deciding what gets backported up front, then create the milesone reactively when the PR is submitted, based on what went in the PR. Having the milestone represent both "proposed work to include" before and "work that was included" after, is hard to manage and error prone.

Sure, that would be my preferred way.

I updated the milestone.

I moved #10451 in to 2.2.2 as well since it appears to be included above. I will work on a release this morning.

Just noting that we missed #10513 for 2.2.2 and that was a fairly serious bug that has broken Dask, for instance. I do agree that patch releases should only contain fixes but I think we should be generous and try to include any fixes that can be cleanly/easily applied, even if they were not strictly regressions. [1] Users want all fixes as fast as possible and do not care about these distinctions.

[1] In this case it was but my point is that splitting hairs over this makes it more likely that regressions will be missed, just like happened here. I'd much rather err on the side of including as many fixes as possible in a patch releases.

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