ipywidgets
ipywidgets copied to clipboard
jupyter-widgets
8.0 Release plans
We are well on our way to an 8.0 release. The milestone is at https://github.com/jupyter-widgets/ipywidgets/milestone/30
I've tagged backwards-incompatible issues/PRs in the 8.0 milestone with a label. I consider those criticial blockers for 8.0. Other issues/PRs could come in an 8.1 if needed since they are not backwards incompatible. Please let us know if you think otherwise, or if you feel an issue/PR has been mislabeled.
~~Since JupyterLab 2.0 (shipping by Feb 2020) requires ipywidgets 8.0 (and not ipywidgets 7.x), I suggest we have a tentative release date of the first or second week of February.~~
Update: We have now released 8.0.0rc2 (5 Aug 2022), which we anticipate being the final RC release. Items still to do:
- [x] Review PRs in the cookiecutter to help authors porting widgets to 8.0:
- [x] https://github.com/jupyter-widgets/widget-cookiecutter/pull/103
- [x] https://github.com/jupyter-widgets/widget-cookiecutter/pull/87
- [x] Note in our migration guide these PRs to help widget authors incorporate their changes as they upgrade their packages
- [x] Add release on GH
- [ ] Add blog post
FYI, JLab just released an RC, so it will probably be about a week until JLab 2.0.
Let's try to do a call (this week) to evaluate the backwards incompatible changes that have already been merged, and whether any need to / should be reverted before 8.0. Given that 8.0 is now a requisite for lab 2.0, we might want to give it a double check, as my initial understanding would be that we would have longer to evaluate the consequences of those changes.
I think some of the more controversial changes (like the rewrite of the data layer) has been postponed past 8.0 given the tight timeline we are on. My guess is that we won't have to revert anything, it's more a question of how much more to merge in.
How about Wednesday at 8am PT?
I have a conflicting meeting at that time, and I'm not sure the reset of @jupyter-widgets/jupyter-widgets has noticed :)
I think some of the more controversial changes (like the rewrite of the data layer) has been postponed past 8.0 given the tight timeline we are on.
I think you are right, at least on the Python/spec side of things (that matters the most w.r.t. stability).
@jupyter-widgets/jupyter-widgets, @zerline, and anyone else interested in participating: I propose that we have a weekly meeting (30-45 minutes) to maintain momentum towards releasing 8.0 in the next month or so.
Anyone interested in joining, please mark times that work for you at http://whenisgood.net/9eftj5s
CC @ibdafna @mbektasbbg - please see the message above about a weekly meeting
I think some of the more controversial changes (like the rewrite of the data layer) has been postponed past 8.0 given the tight timeline we are on. My guess is that we won't have to revert anything, it's more a question of how much more to merge in.
For completeness - we released a new version of the jlab widget manager compatible with ipywidgets 7.5 and jupyterlab 2.0, so the time pressure is off for shipping ipywidgets 8.0 with jlab 2.0.
Note that if we were to have it bi-weekly, we could have it in the same slot as the Voila call (alternating weeks).
Since there are no times that work for the three people responding to the survey, let's try @vidartf's idea, and try it in the Voila meeting timeslot in the off-weeks from the voila meeting.
Looking at https://voila-dashboards.github.io/, that would mean the first meeting would be 9:30-10:30 PDT Monday, March 30, 2020.
Would that work better for @zerline or @jupyter-widgets/jupyter-widgets?
Note that the voila meet is specified in central EU timezone that switched to DST this weekend, but @jasongrout specified the call in PDT above, so I'm confused what time we are operating with here.
I'm happy to leave the call anchored to a EU timezone. What time is the Voila call if we anchor to Europe? I can meet either now or in an hour, which I think is what the two possibilities are.
We can use the Voila Weekly Whereby room for these calls: https://whereby.com/voila-weekly
Time is 17:30 CET: https://voila-dashboards.github.io
@jupyter-widgets/jupyter-widgets and others that want to help with ipywidgets 8: @ianhi is setting up a meeting for us to restart planning for ipywidgets 8:
If you're interesting joining a ipywidgets meeting please fill out this whenisgood http://whenisgood.net/hbbqe7h
The broad goals of this meeting are to talk briefly about high level goals and to figure out the state of an 8.0 release.
See https://hackmd.io/V2rlXrDQRzKH3lyTgh0ZIw for full details.
As @willingc requested: FYI, @willingc, @mseal, @captainsafia - we are arranging an ipywidgets 8.0 planning meeting soon. See the comment just above this for details.
Hi all,
Following up on this. Unfortunately there was no single time that worked for everyone. So the time that was settled on (https://gitter.im/jupyter-widgets/Lobby?at=600ae1028816425540e2ac46) is: Tuesday, Feb 2nd, at 3 pm UTC (10 am EST)
@jasongrout how do links for things like this get generated? Zoom links? Generated by whoever?
@jasongrout @ianhi It would be great if we can record this meeting for those unable to attend. Thanks for organizing.
Thanks all for organizing.
Here is a link to easily get the time and date of the meeting for your time zone: Tuesday, Feb 2nd, at 3 pm UTC
(also added to the hackmd)
Thanks everyone who participated in the meeting today. It was really great to see everyone.
I've added notes at a new team-compass repo: https://github.com/jupyter-widgets/team-compass/issues/1
I've released ipywidgets 8.0.0a3, widgetsnbextension 4.0.0a1, jupyterlab_widgets 2.0.0a0. This works in JupyterLab 3.
The current plan is a first beta release of 8.0 next week, with an RC phase starting around the beginning of December. If no major disruptions are encountered, the plan is then to do a final release of 8.0 in early/mid January.
Due to some delays in the release schedules, the RC phase seems likely to not start until early January, with a subsequent shift of the final release to late January.
I'm testing RPM packages in Fedora with IPython 8.0.0rc1, ipywidgets 8.0.0b1, and widgetsnbextension 4.0.0b1 and so far I've discovered only one minor test incompatibility in nbclient, reported: https://github.com/jupyter/nbclient/issues/191
Commented on the nbclient issue: https://github.com/jupyter/nbclient/issues/191#issuecomment-1008651338
Any news on final RC/release plans ? The target seems to have shifted somehow Thanks for your work !
It was initially delayed by a last-minute request to resolve #3111. A suggested solve to that issue was merged in #3195. The final piece currently missing for the RC is the attempt to make that implementation be backwards compatible on the protocol level (i.e. avoid needing to bump the widget messaging protocol to a new major version) in #3394.