fix: Resolve pa.scalar overload overlaps

[dangotbanned]

• Avoids 18 warnings from pyright
• very similar to fix: Avoid chunked_array overlapping overloads #183
• MonthDayNano was a bit tricky, but is based on https://github.com/python/typeshed/blob/a045be8ed62492952595661118af22554c7cbca4/stdlib/_typeshed/__init__.pyi#L316-L333
  • Had to rewrite manually to avoid overload overlaps on CollectionValue
  • Has all the properties/members returns by inspect.getmembers
  • Signature via inspect.signature

scalar.pyi:281:5 - error: Overload 3 for "scalar" overlaps overload 4 and returns an incompatible type (reportOverlappingOverload)
scalar.pyi:281:5 - error: Overload 3 for "scalar" overlaps overload 5 and returns an incompatible type (reportOverlappingOverload)
scalar.pyi:297:5 - error: Overload 7 for "scalar" overlaps overload 8 and returns an incompatible type (reportOverlappingOverload)
scalar.pyi:313:5 - error: Overload 11 for "scalar" overlaps overload 16 and returns an incompatible type (reportOverlappingOverload)
scalar.pyi:313:5 - error: Overload 11 for "scalar" overlaps overload 17 and returns an incompatible type (reportOverlappingOverload)
scalar.pyi:313:5 - error: Overload 11 for "scalar" overlaps overload 26 and returns an incompatible type (reportOverlappingOverload)
scalar.pyi:327:5 - error: Overload 13 for "scalar" overlaps overload 14 and returns an incompatible type (reportOverlappingOverload)
scalar.pyi:327:5 - error: Overload 13 for "scalar" overlaps overload 15 and returns an incompatible type (reportOverlappingOverload)
scalar.pyi:327:5 - error: Overload 13 for "scalar" overlaps overload 18 and returns an incompatible type (reportOverlappingOverload)
scalar.pyi:327:5 - error: Overload 13 for "scalar" overlaps overload 19 and returns an incompatible type (reportOverlappingOverload)
scalar.pyi:327:5 - error: Overload 13 for "scalar" overlaps overload 20 and returns an incompatible type (reportOverlappingOverload)
scalar.pyi:327:5 - error: Overload 13 for "scalar" overlaps overload 21 and returns an incompatible type (reportOverlappingOverload)
scalar.pyi:327:5 - error: Overload 13 for "scalar" overlaps overload 22 and returns an incompatible type (reportOverlappingOverload)
scalar.pyi:327:5 - error: Overload 13 for "scalar" overlaps overload 23 and returns an incompatible type (reportOverlappingOverload)
scalar.pyi:327:5 - error: Overload 13 for "scalar" overlaps overload 24 and returns an incompatible type (reportOverlappingOverload)
scalar.pyi:327:5 - error: Overload 13 for "scalar" overlaps overload 25 and returns an incompatible type (reportOverlappingOverload)
scalar.pyi:369:5 - error: Overload 19 for "scalar" will never be used because its parameters overlap overload 18 (reportOverlappingOverload)
scalar.pyi:376:5 - error: Overload 20 for "scalar" will never be used because its parameters overlap overload 18 (reportOverlappingOverload)

Example of flaky false-negative

Mentioned in

• refactor: Simplify types overloads #219 (comment)
• fix: Resolve all array.pyi errors #209 (comment)

Loading a module for the first time fails to match an overload

Editing the file (e.g. adding a space) "fixes" the issue

[zen-xu]

Can you give an example?

[dangotbanned]

Can you give an example?

@zen-xu I'm not sure I understand correctly, so apologies if this is the wrong part.

Example 1

The example I gave comes from https://github.com/narwhals-dev/narwhals/blob/9d44834321c6cdeecb902f3a41742669a2156cc5/narwhals/_arrow/series.py#L1100

But as I mentioned in (#219 (comment)), this issue can show up anywhere pa.scalar or pa.array are used without a type argument provided.

Example in description

Example of flaky false-negative

Mentioned in

• refactor: Simplify types overloads #219 (comment)
• fix: Resolve all array.pyi errors #209 (comment)

Loading a module for the first time fails to match an overload

Editing the file (e.g. adding a space) "fixes" the issue

Example 2

This was the most minimal repro I could do for the performance issue (ignoring my initial typo 😄)

2025-04-22.10-23-20.-.scalar.mp4

Specs

I've not had performance issues from stubs before - but just to verify - I don't think I'm pushing my PC too far here

• CPU: AMD Ryzen 9 3900X 12-Core (24-thread) 3.80 GHz
• RAM: DDR4 32.0 GB
• Storage: Samsung SSD 980 PRO 1TB

[dangotbanned]

FYI, if any merge conflicts pop up in the future - this change is most likely the source.
I needed to bump ruff to get a fix in CI that was required by pre-commit.

• (bccc217)
• (076cf54)

[zen-xu]

Example 2 did not reproduce on my Mac

• pylance: 2024.8.1
• vscode: 1.96.2
• pyright type checking mode: basic

[dangotbanned]

Example 2 did not reproduce on my Mac

* pylance: 2024.8.1

* vscode: 1.96.2

* pyright type checking mode: basic

@zen-xu could you try a more recent pylance release?

The version you're using seems to be from 8 months ago - which would be before I started using pyarrow-stubs

• https://github.com/microsoft/pylance-release/releases/tag/2025.4.1

[dangotbanned]

Oh for the second example, it is important to have these settings enabled

"python.analysis.inlayHints.functionReturnTypes": true,
"python.analysis.inlayHints.variableTypes": true,

[zen-xu]

I think #233 has resolved this problem