`bpy.data.libraries` not sortable

[drsybren]

This code causes a Pylance error:

import bpy

libs = sorted(bpy.data.libraries, key=lambda lib: lib.filepath)

The error:

No overloads for "sorted" match the provided arguments
Pylance reportCallIssue

Argument of type "Unknown | Sequence[Library] | Mapping[str, Library] | bpy_prop_collection" cannot be assigned to parameter "iterable" of type "Iterable[_T@sorted]" in function "sorted"
  Type "Unknown | Sequence[Library] | Mapping[str, Library] | bpy_prop_collection" is not assignable to type "Iterable[_T@sorted]"
    "bpy_prop_collection" is incompatible with protocol "Iterable[_T@sorted]"
      "__iter__" is not present

Pylance: reportArgumentType

The code runs fine, though.

[michaelgold]

The stubgen generates this type for bpy.data.libraries

libraries: typing.Union[BlendDataLibraries, typing.Sequence[Library], typing.Mapping[str, Library], bpy_prop_collection] = ...

It's expecting a union of types (some of which I believe are not iterable). This may be the intended result, and could explain why it works at runtime: the runtime type of libs was typing.Sequence[Library]

I was able to avoid a Pylance error by explicitly casting libs to a Sequence[Library]

import bpy
import bpy.types
from typing import Sequence, cast

libs = sorted(
    cast(Sequence[bpy.types.Library], bpy.data.libraries), key=lambda lib: lib.filepath
)

[drsybren]

I don't understand why this is closed. IMO the stubs should be representative of Blender's functionality, and thus the fact that bpy.data.libraries can be used as an iterable should be expressed by the stubs.

[michaelgold]

You're right. I tracked it down further to what I believe is the root cause (the stubgen is not actually looking at the c types for this class)
https://projects.blender.org/blender/blender/issues/145140