1.16 Release Planning

I am planning to release 1.16 in the coming weeks.

If there are any PRs you would like to ensure are included, please mention here.

This issue will be updated once the release branch is cut.

Note there's a crash on mypy master we may want to address #18764

Update: fixed by Ivan

FWIW worth including two low-risk crash fixes in the release:

What's the current status here? Has the release branch been cut already?

If not, I'd suggest to do it after the current typeshed sync #18880. This is the last commit with support for type checking Python 3.8. For the next typeshed sync, we'll need to drop support for --python-version 3.8 completely. Might be great to get one last release out before that.

If possible, we should include #18906 in the new release. It fixes high pri daemon bug

If not, I'd suggest to do it after the current typeshed sync #18880. This is the last commit with support for type checking Python 3.8. For the next typeshed sync, we'll need to drop support for --python-version 3.8 completely. Might be great to get one last release out before that.

Why not just drop 3.8? It has been EOL for over 6 months now

Why not just drop 3.8? It has been EOL for over 6 months now

We already dropped 3.8. It's just that mypy still supports checking 3.8 code with --python-version 3.8 for now. This will become impossible with the next typeshed update #18930.

Hmm that's too bad. Because #18930 contains ctypes improvements (python/typeshed#13777 and python/typeshed#13760) that matter to numpy.

Hmm that's too bad. Because #18930 contains ctypes improvements (python/typeshed#13777 and python/typeshed#13760) that matter to numpy.

I'd be fine with dropping it now as well. The idea just was to do one last release before then. The release branch could have even been cut already. I just don't know.

Will it be possible to have #15043 in 1.16? If not, can I please ask for some feedback on the issue as to what else is required before it's merged?

For NumPy #18997 is a big problem. We won't be able to release numpy/numtype (a rework of numpy's typing stubs with new (backwards-incompatible) features like shape-typing).

The plan is to publish an the initial NumType release in one or two months. So if this isn't fixed in mypy 1.16, our only option would be to drop support for mypy. I really don't want that.

@meshy

@meshy There was a recent change that likely will make ancestors included in seen modules, so I think your PR can likely be simplified. That said however, realistically I think it likely won't make it into 1.16, but I promise I will get back to it as soon as I am done with other 1.16-related work (I was bothered by this issue myself, so I am interested in it being fixed).

@jorenham I strongly suggest you drop this "or otherwise" attitude. If you think it will persuade someone, then I can tell you it will be the opposite.

I strongly suggest you drop this "or otherwise" attitude. If you think it will persuade someone, then I can tell you it will be the opposite.

Oh woops I see how it can be interpreted that way, and I did not mean it as a threat or something. I was just trying to say that I consider it to be very important that we keep on supporting mypy, and that I'm therefore willing to help in any way I can in solving this issue.

I opened #19001 to address #18997. Hopefully we can include it in 1.16.

Prepared a revert of #19139 which we can merge if the regression is tricky or slow to fix:

Revert "Infer correct types with overloads of Type[Guard | Is] #19161

Status of regressions:

[1.16 regression] determining exhaustive match on an enum with type members #19151 was fixed.
[1.16 regression] narrowing on is_dataclass #19139 has an open revert PR as a workaround. I will probably merge the revert tomorrow if there isn't a fix available.
[1.16 regression] Incorrect inferred type when using a classmethod from a different class as an attribute #19146 is a regression but since 1.16 makes the behavior of mypy more consistent (i.e. it's a side effect of fixing an inconsistency), we may not fix it.
[1.16 regression] failure to exhaustively narrow type var bounded by union with value pattern match #19159 is a regression but it doesn't seem urgent to fix, since it only triggers under a rare set of conditions.

Unless there are concerns or new reported regressions, I think we are on track to finally get the release out tomorrow (Thu).

Add changelog for 1.16 (#19138)

Mypy 1.16 is out: https://mypy-lang.blogspot.com/2025/05/mypy-116-released.html

Mypy 1.16 is out: mypy-lang.blogspot.com/2025/05/mypy-116-released.html

Great! Thanks for your hard work and for keeping us in the loop :)

Seems to be some kind of regression here:
https://github.com/aio-libs/aiohttp/actions/runs/15351081200/job/43199239374?pr=11093

As far as I can see, the type should get inferred automatically as Any:
https://github.com/aio-libs/aiohttp/blob/2a8ff49426929569193cb77c67d6682e8b008bce/aiohttp/tracing.py#L48-L58
https://github.com/python/typeshed/blob/a92ecca05417b45d02d9396e4892ca67e80e2891/stdlib/unittest/mock.pyi#L190

If that's not a known/obvious issue, I'll try and look at it in more depth later and open an issue.

Otherwise, this release has allowed us to remove dozens of type ignore comments across our projects, so has been looking pretty good.

@Dreamsorcerer

@Dreamsorcerer I don't think we have an issue about this, and it doesn't look like something obvious. I think better to open a new issue.

Add changelog for 1.16 (python#19138)

Three fixes for regressions in 1.16 (well, technically only the first one is a real regression, other two just happen to occur more often for unrelated reasons):

Also a fix a (corner case) crash introduced in 1.16:

Fix crash on invalid property inside its own body #19208

Regressions fixed:

(crash) [1.16 regression] crashes on invalid and self-referential setter #19205
mypyc 1.16 regression: x86_64-cpython-3{9,10,11,12,13} error: ‘FOO_finalize’ undeclared here (not in a function); did you mean ‘FOO_init’? #19175
[1.16] Regression in inferring type via unittest.mock.Mock #19184
[1.16 makes more common] Inference of Any against optional type infers Never #8829
[1.16 regression] assigning a lambda to subclass w/ generics #19165
[1.16 regression] determining exhaustive match on an enum with type members #19151
(via revert) [1.16 regression] narrowing on is_dataclass #19139
(crash) [1.16 regression] Internal error: must never apply partial type #19213
[1.16 regression] narrowing a typevar via comparison to None #19166
[1.16 regression] a cast is both needed and redundant at the same time #19170
[1.16 regression] Incorrect inferred type when using a classmethod from a different class as an attribute #19146
[1.16 regression] Relative order of property methods changes mypy output #19224
[1.16 regression] duplicated note for incompatible type #19240
[1.16 regression] no longer recognizes enum singleton as only possible case #19271

Unresolved regressions:

Since there were quite a few regressions, I think it's worth making a point release.

#19118 is quite a big change and I think it can wait until 1.17, even though it fixes regressions.

If nobody else beats me to it, I'll probably make the release early next week.

These seem safe to cherry pick:

Fix crash on invalid property inside its own body #19208 for [1.16 regression] crashes on invalid and self-referential setter #19205
[mypyc] Fixing condition for handling user-defined __del__ #19188 for mypyc 1.16 regression: x86_64-cpython-3{9,10,11,12,13} error: ‘FOO_finalize’ undeclared here (not in a function); did you mean ‘FOO_init’? #19175
Fix crash on partial type used as context #19216 for [1.16 regression] Internal error: must never apply partial type #19213
Revert "Infer correct types with overloads of Type[Guard | Is] #19161 for [1.16 regression] narrowing on is_dataclass #19139
Tighten metaclass __call__ handling in protocols #19191 for [1.16] Regression in inferring type via unittest.mock.Mock #19184
Allow enum members to have type objects as values #19160 for [1.16 regression] determining exhaustive match on an enum with type members #19151
Handle assignment of bound methods in class bodies #19233 for [1.16 regression] Incorrect inferred type when using a classmethod from a different class as an attribute #19146
Fix properties with setters after deleters #19248 for [1.16 regression] Relative order of property methods changes mypy output #19224
Single underscore is not a sunder #19273 for [1.16 regression] no longer recognizes enum singleton as only possible case #19271

These seem less safe to cherry pick:

Regardless of patch release, it could be good to roll pretty quickly into a 1.17. There are already some substantial unreleased changes and people will appreciate speed fix from tyralla's PR as well

Mypy 1.16.1 is out with the safe fixes mentioned above cherry picked (a few of them were already cherry picked to 1.16.0). I'll prepare a short blog post tomorrow.

Navigation Menu

Repository navigation

1.16 Release Planning #18739

16 remaining items

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

1.16 Release Planning #18739

Description

Activity

hauntsaninja commented on Mar 7, 2025

ilevkivskyi commented on Mar 31, 2025

cdce8p commented on Apr 4, 2025

svalentin commented on Apr 16, 2025

jorenham commented on Apr 16, 2025

cdce8p commented on Apr 16, 2025

jorenham commented on Apr 16, 2025

cdce8p commented on Apr 16, 2025

meshy commented on Apr 28, 2025

jorenham commented on Apr 29, 2025

ilevkivskyi commented on Apr 29, 2025

jorenham commented on Apr 29, 2025

jorenham commented on Apr 29, 2025

16 remaining items

JukkaL commented on May 28, 2025

JukkaL commented on May 28, 2025

JukkaL commented on May 29, 2025

jorenham commented on May 29, 2025

Dreamsorcerer commented on May 30, 2025

ilevkivskyi commented on May 30, 2025

ilevkivskyi commented on Jun 3, 2025

hauntsaninja commented on Jun 3, 2025

JukkaL commented on Jun 6, 2025

hauntsaninja commented on Jun 6, 2025

JukkaL commented on Jun 16, 2025

Metadata

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

Issue actions