Fix Literal bug with typing-extension==4.6.0#5826
Conversation
7f0a4d5 to
93ef367
Compare
AlexWaygood
left a comment
There was a problem hiding this comment.
I'm not familiar with pydantic internals, but this looks like a reasonable fix to me!
Co-authored-by: Alex Waygood <Alex.Waygood@Gmail.com>
Sure. Thanks! |
|
Will this fix be delivered as Pydantic 1.10.8 immediately? If yes, I will wait and not implement any workarounds. |
989eecb to
aab48c9
Compare
|
I wonder if it might not be a good idea to introduce an upper bound ( The same applies to the master branch of course. |
|
Honestly, I feel like this situation illustrates the inadequacies of SemVer more than anything else |
agreed.
No. There are many articles out there about why including upper bounds in package dependencies is almost always a bad idea.
We'll do our best to get it out asap. |
|
Although I don't feel like |
|
No problem @AlexWaygood, these things happen. It's the responsibility of app developers to pin dependencies. CF build failures is a random download issue, see my angry tweet. |
| if type_ is none_type: | ||
| return True | ||
| return False | ||
| return type_ in NONE_TYPES |
There was a problem hiding this comment.
This looks weird enough that we might have done it for some strange reason, can we confirm this was just a code smell?
There was a problem hiding this comment.
The existing approach(checking with is) does not work in Python 3.9 and typing-extension==4.6.0. That's why I've changed it.
Note: the new approach is also Ok with typing-extension==4.5.0. So, it might be a wrong implementation before
There was a problem hiding this comment.
thanks @hramezani.
Can we uprev the version we test with?
|
This should be fixed now in the just-released 1.10.8, which should be available through PyPI shortly. |
Done |
Fixes #5821