[ty] Prefer declared type of generic classes

[ibraheemdev]

Summary

Extends #20927 to generic call inference, preferring the declared type of generic classes:

def f[T](x: T) -> list[T]:
    return [x]

x: list[Any] = f(1)
reveal_type(x)  # before: list[Literal[1]], after: list[Any]

Part of astral-sh/ty#136.

[github-actions]

Diagnostic diff on typing conformance tests

No changes detected when running ty on typing conformance tests ✅

[github-actions]

mypy_primer results

Changes were detected when running on open source projects

dulwich (https://github.com/dulwich/dulwich)
- dulwich/pack.py:1703:83: warning[unused-ignore-comment] Unused blanket `type: ignore` directive
- Found 172 diagnostics
+ Found 171 diagnostics

pydantic (https://github.com/pydantic/pydantic)
- pydantic/_internal/_validators.py:161:16: error[invalid-return-type] Return type does not match returned value: expected `Pattern[bytes]`, found `Pattern[str | bytes]`
+ pydantic/_internal/_validators.py:161:16: error[invalid-return-type] Return type does not match returned value: expected `Pattern[bytes]`, found `Pattern[bytes | str]`

Expression (https://github.com/cognitedata/Expression)
- tests/test_result.py:491:21: error[invalid-argument-type] Argument to bound method `or_else` is incorrect: Expected `Result[Literal[42], Any]`, found `Result[Literal[0], Any]`
- tests/test_result.py:509:21: error[invalid-argument-type] Argument to bound method `or_else` is incorrect: Expected `Result[Any, Literal["original error"]]`, found `Result[Any, Literal["new error"]]`
- Found 201 diagnostics
+ Found 199 diagnostics

sphinx (https://github.com/sphinx-doc/sphinx)
- sphinx/domains/python/__init__.py:672:42: error[index-out-of-bounds] Index 0 is out of bounds for string `Literal[""]` with length 0
- Found 530 diagnostics
+ Found 529 diagnostics

jax (https://github.com/google/jax)
- jax/_src/interpreters/mlir.py:1484:5: error[invalid-assignment] Object of type `OrderedDict[str, Unknown]` is not assignable to attribute `_tokens` of type `OrderedDict[Effect, Unknown]`
- Found 2544 diagnostics
+ Found 2543 diagnostics

rotki (https://github.com/rotki/rotki)
+ rotkehlchen/exchanges/coinbase.py:524:20: error[invalid-assignment] Object of type `dict[Unknown, Unknown]` is not assignable to `defaultdict[str, list[dict[Unknown, Unknown]]]`
- Found 1684 diagnostics
+ Found 1685 diagnostics

core (https://github.com/home-assistant/core)
+ homeassistant/helpers/service.py:913:9: error[invalid-assignment] Object of type `object` is not assignable to `ServiceResponse`
- Found 14272 diagnostics
+ Found 14273 diagnostics

Memory usage changes were detected when running on open source projects

prefect (https://github.com/PrefectHQ/prefect)
-     struct metadata = ~33MB
+     struct metadata = ~35MB

[github-actions]

ecosystem-analyzer results

Lint rule	Added	Removed	Changed
invalid-assignment	2	1	0
invalid-argument-type	0	2	0
index-out-of-bounds	0	1	0
invalid-return-type	0	0	1
unused-ignore-comment	0	1	0
Total	2	5	1

Full report with detailed diff (timing results)

[ibraheemdev]

I'm not sure if we should restrict this change to invariant generics. It seems reasonable to use frozendict[str, Any] as an implicit typed dict, so preferring the Any there might still be meaningful (even thought it feels strictly worse).

[carljm]

@ibraheemdev Is this blocked? What's keeping it in draft?

[codspeed-hq]

CodSpeed Performance Report

Merging #21070 will not alter performance

_{Comparing ibraheem/declared-generic-type (2f4dcbf) with main (1d111c8)}

Summary

✅ 22 untouched
⏩ 30 skipped¹

Footnotes

1. 30 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[ibraheemdev]

The ecosystem changes all seem correct.

[ibraheemdev]

Superceded by #21210.