Conformance tests and spec change for generic type erasure #1589
Conversation
| @@ -333,15 +333,15 @@ argument(s) is substituted. Otherwise, ``Any`` is assumed. Example:: | |||
|
|
|||
| class Node(Generic[T]): | |||
There was a problem hiding this comment.
The paragraph above could probably be improved. For instance, without the change I make here, pyright would infer Node() to be Node[None], and I think that should be legal.
That said, handling of default values for generics in general is underspecified (mypy handles them poorly, but in part because it wasn't clear what the behaviour should be python/mypy#3737)
There was a problem hiding this comment.
When this part of the spec was written, mypy defaulted to "implicit-optional", which would give you the same effect as adding | None. Since then, the spec was updated to recommend against "implicit-optional", but this code sample wasn't updated. I'd be in favor of just updating the spec to fix the sample.
| p.x # Ok (evaluates to None) | ||
| Node[int]().x # Ok (evaluates to None) | ||
| p.x # Ok (evaluates to int) | ||
| Node[int]().x # Ok (evaluates to int) | ||
| p.x = 1 # Ok, but assigning to instance attribute |
There was a problem hiding this comment.
These just seem straight up incorrect to me
There was a problem hiding this comment.
Yes, I agree. Looks like a spec bug.
|
@hauntsaninja, is this PR ready to merge, or do you have additional work to do on it? |
|
This is ready to merge, I just wanted to leave open for a few days since I change the meaning of the spec. (And I still owe tests for "Abstract generic types" and "Type variables with an upper bound") |
No description provided.