Erik De Bonte

That matches the behavior of unquoted generic types. 

I disagree. The type var should be a different color to make it clear that it is a distinct identifier that supports, for example, go to def. I'm indifferent about...

I see. The brackets are colored as string characters. I agree, that's weird. I'm not convinced that the brackets should be the same color as the type name though.

@tiangolo, if you have any thoughts on how Pylance could improve the FastAPI development experience, we'd love to hear them!

Related: https://github.com/microsoft/pyrx/issues/3380, https://github.com/microsoft/pyrx/issues/3381, https://github.com/microsoft/pyrx/issues/3382

@pawamoy, yes, pyrx is a private repository used by the Pylance team. Sorry for the confusion. I just wanted to link together our public issues and internal issues related to...

Hi @tiangolo! When we add experimental support for a draft PEP, it's generally with the goal of helping the community make a decision on whether that PEP should be accepted....

> I guess the decision would be similar to adding support or not for Numpy docstrings, scaling the user base (and estimated user growth) to FastAPI instead of Numpy. @tiangolo,...

I am also able to repro this. Seems that it regressed at some point. Maybe due to the URI refactoring? It looks like the tests for this scenario were removed...

I've been trying off and on over the past couple weeks to reproduce this issue, but no luck. I've tried: - Using black/ruff to cause edits as [suggested by @s-banach](https://github.com/microsoft/pylance-release/issues/4295#issuecomment-1529714441)...