Repository navigation
Release 0.930 planning #11700
Description
I'm planning another release fairly soon after 0.920 (#11158) is out to catch up with the GitHub master branch. If everything goes as planned, I'll cut the release branch around Tue 14 Dec. Please post here any PRs that you'd like to include in the release and that are not yet in master (or in the release branch after I've created it).
Note: Please do not sync typeshed until I've created the release branch. There are already some typeshed regressions that likely need to be resolved. If you want to include some specific typeshed changes, I can potentially cherry-pick them.
Activity
JukkaL commentedon Dec 10, 2021
Regression that is a release blocker: #11699 (_order_in enum)JukkaL commentedon Dec 10, 2021
Another regression:
#11701(Literal[True, False]false positives)JukkaL commentedon Dec 10, 2021
Another regression that might be a release blocker (but it might be very rare and not super important to fix):
#11702(enum inheritance false positive)JukkaL commentedon Dec 10, 2021
Yet another regression that looks like a release blocker:
#11704(need type annotation with contextmanager)JukkaL commentedon Dec 10, 2021
One more regression: #11705 ("and" type inference with strict optional disabled)
hauntsaninja commentedon Dec 11, 2021
Currently unmerged PRs I'd like included:
#11681(low risk, fixes a crash)#11708(low risk, removes an optimisation that resulted in a regression reported multiple times)#11709(fixes #11701 as mentioned above)If possible, enough of #10191 that we no longer crash on match statementsCurious if @sobolevn has thoughts on any PRs to prioritise
sobolevn commentedon Dec 11, 2021
Direct blockers:
Enumand some special props to be non-final, refs #11699 #11713Also a couple of bug fixes that can potentially be merged:
selfare handled properly #11366ifexpr #11128__hash__method to@dataclasses.dataclass, refs #11463 #11496Type[None].__bool__is a thing, refs #11539 #11551 and Nowtype(None)is more consistent, refs #11550 #11552Zac-HD commentedon Dec 12, 2021
#11718keeps biting me, and should be pretty easy to merge.JukkaL commentedon Dec 14, 2021
Two planned reverts to avoid regressions:
#11741(Python 2 division),#11743(Python 2 raise), unless there's a forward fix very soon.Some cherry-picks from typeshed to avoid regressions:
#11739sobolevn commentedon Dec 14, 2021
Sorry, I won't be able to submit a fix before the weekend. If this is fine 🙂
Or you can always keep this out of scope of the release, these two feature are not very important.
JukkaL commentedon Dec 14, 2021
No worries, I reverted the PRs now so that they won't block the release.
cdce8p commentedon Dec 15, 2021
Don't know if there is still some time left but I believe #10712 would be a good feature to have.
hauntsaninja commentedon Dec 16, 2021
I'd like this typeshed PR to be included:
python/typeshed#6591(fixes a regression shipped with 0.920)18 remaining items
JukkaL commentedon Jan 4, 2022
I'm planning to publish a 0.931 bugfix release later this week (current estimate is Wed or Thu). Thank you everybody who contributed fixes to regressions, and thank you @hauntsaninja for the useful summary!
In addition to the PRs mentioned above, this also fixes a regression:
JukkaL commentedon Jan 5, 2022
Here are the changes I've cherry-picked to the
release-0.930branch:excludein TOML files #11828__slots__regression in 0.930, refs #11821 #11824__module__and similar non-finalEnumnames #11823I'm planning to make the 0.931 release with these changes tomorrow, unless something new comes up.
Additional things that might be worth fixing (haven't looked into these in much detail):
NoReturnno longer works as a type annotation #11903A5rocks commentedon Jan 6, 2022
#11913 is also a regression it seems.
hauntsaninja commentedon Jan 6, 2022
#11912fixes a minor regression in 0.930 (caused by a changed in typeshed stubs)#11918and#11863are not a regression, but would make typeshed folks happy. We indicated we'd be willing to issue point releases for this in python/typeshed#4820 (comment)JukkaL commentedon Jan 6, 2022
I cherry-picked these additional fixes:
type[T]in stubs #11863NoReturntype aliases #11912Since there are still a few regressions without a fix, I will wait until tomorrow (Friday) before making the release. If important regressions are still open tomorrow, I can make a 0.932 release once we have the remaining fixes.
JukkaL commentedon Jan 6, 2022
Another potential regression:
hauntsaninja commentedon Jan 7, 2022
#11717 was merged, but not backported
I believe right now #11924 is the only open PR that fixes a regression
sobolevn commentedon Jan 7, 2022
PR with
stubtestbugfix: #11931This would be helpful for
typeshedJukkaL commentedon Jan 7, 2022
More cherry-picks:
Nonehandling with--no-strict-optional, refs #11705 #11717ErasedTypeandcovers_at_runtime#11924JukkaL commentedon Jan 7, 2022
I'm planning to include this PR in the release:
JukkaL commentedon Jan 7, 2022
Two final cherry-picks for 0.931:
We should be ready for the release soon, once the wheel build completes. I'll need to prepare a short release blog post as well.
JukkaL commentedon Jan 7, 2022
...and 0.931 is out: http://mypy-lang.blogspot.com/2022/01/mypy-0931-released.html
Handle `raise Exception(), None` on Python2.7 (#11786)
Use `__truediv__` for python2 with `__future__` import, refs #11740 (#…