python / python/mypy

Mypy is still too eager about outer context

Open
#5,874 3 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

bug false-positive priority-1-normal topic-type-variables topic-union-types
Dominant language
Python
Stars
20.6k
Forks
3.3k
PR merge metrics
PR metrics pending

Description

Consider this example:

T = TypeVar('T')
S = TypeVar('S')

def add(x: List[T], y: List[S]) -> List[Union[T, S]]: ...

a: List[int]
b: List[str]

e: List[Union[int, str]] = add(a, b)

It fails on the last line with the following (I skip invariance notes):

error: Argument 1 to "add" has incompatible type "List[int]"; expected "List[Union[int, str]]"
error: Argument 2 to "add" has incompatible type "List[str]"; expected "List[Union[int, str]]"

If I remove the annotation for e, everything works fine.

When previously working on the external context issues, I found that I don't like how constrains for unions against unions are inferred (not that they are wrong, just counterintuitive). For example, Union[T, S] <: Union[str, int] infers T < : Union[str, int] and S <: Union[str, int], which is technically fine, but leads to errors like above.

This blocks https://github.com/python/mypy/pull/5873

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reproducing the annotated add(a, b) example with mypy and read the discussion around the referenced pull request 5873. Determine the intended constraint inference for unions against unions, then verify that the example succeeds without introducing regressions.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
compilers
Issue type
Bug
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.