Is it possible for 3rd party libraries to work with environment Python version when pythonVersion is set?

[Azureblade3808]

I've been working on a project which is getting used mainly with Python 3.8 (to support Windows 7) and Python 3.13. The Pyright option pythonVersion is set to "3.8" so that I won't accidentally introduce some stuff which is incompatiable with Python 3.8.

The problem is, when I switch the developing environment to one based on Python 3.13, numpy.typing.NDArray becomes unusable.

Version of numpy in that environment is 2.2.0, and the declaration of NDArray is roughly like --

from typing import TypeAlias

NDArray: TypeAlias = ...

As you may have noticed, typing.TypeAlias is not available in Python 3.8, so NDArray is regarded as a variable instead of a type.

I see that different pythonVersion settings can be applied to different part in the project, but what about 3rd party libraries?

[erictraut]

No, it's not possible to set the pythonVersion to be different for a third-party library. Pyright doesn't analyze third-party libraries separately from your files. When you import a symbol from a third-party library, its type is evaluated using the settings from the file that imports it. After all, the code in your file will be executed in the same environment as the library code that it's importing.

As you've noted, you can specify a different pythonVersion setting for different subdirectories in your project using the "execution environment" configuration support.

If you're developing code that must remain compatible with Python 3.8, I recommend using a 3.8 environment for your development. This will keep the development and production environments aligned. If you switch to a newer version, you're likely to run into problems like the one you're experiencing.