Repository navigation
Conversation
|
Hmm could you spell out the situation in which this bound is useful? |
|
According to mypy_primer, this change doesn't affect type check results on a corpus of open source code. ✅ |
Hmm, it depends. The user has to pin an old mypy version without pinning mypy_extensions while using the old Tbh I'm not sure myself how common that is. For Ideally everybody has stopped the old What do you think? In the end it's just an edge case anyway, so I'd be also be fine if we don't do it. |
|
Hm, that sounds niche, and I'm not even sure the mypy version even matters here (if the user is pinning the mypy version, they're not going to pick up this PR). I think we should just roll forward, if we want to avoid breaking things here then we should just not make the breaking change in mypy_extensions in the first place. Going to close, but lmk if I got it wrong! |
Yeah, it is. I debated it myself if I should even open it. Was slightly in favor, so I did.
Yes, this would have only applied to mypy versions released between now and
I'd personally be completely fine with removing it after a deprecation period. The sense I've got so far though is that breaking changes in mypy are heavily discouraged. It would just be awesome if we could get rid of
👍🏻 Works for me. |
mypy_extensions.TypedDictandmypy_extensions.NoReturnwere recently deprecated and will be removed eventually. Add an upper bound now so we don't break old code using these types in the future, after they are removed frommypy_extensions.