Type argument lost when union of collections is passed to generic function
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 20.6k
- Forks
- 3.3k
- PR merge metrics
- PR metrics pending
Description
Bug Report
Passing a union of collections to a generic function that should preserve the type argument erases it instead. (Or should I say it consolidates it to object?)
To Reproduce
from typing import Union
x: list[int] | list[str]
y: Union[set[int], set[str]]
reveal_type(set(x))
reveal_type(list(y))
Expected Behavior
Revealed type is "Union[builtins.set[builtins.int], builtins.set[builtins.str]]"
Revealed type is "Union[builtins.list[builtins.int], builtins.list[builtins.str]]"
I hope this is reasonable. I mean, the type argument is either one or the other and nothing else.
Actual Behavior
Revealed type is "builtins.set[builtins.object]"
Revealed type is "builtins.list[builtins.object]"
Environment
Python 3.10.8 on Linux; mypy --strict version 0.982
More examples
Clearly mypy is capable of preserving the type argument, when there is no union of iterables passed to set:
reveal_type(set([1, 2])) # Revealed type is "builtins.set[builtins.int]"
It also does not seem to have anything to do with the wrong typeshed stubs, as the same problem occurs with custom generic functions:
from collections.abc import Iterable
from typing import TypeVar
T = TypeVar("T")
def f(i: Iterable[T]) -> T:
...
z: list[int] | list[str]
reveal_type(f(z)) # Revealed type is "builtins.object"
The following gives me an error:
def my_set(a: list[int] | list[str]) -> set[int] | set[str]:
return set(a) # Incompatible return value type (got "Set[object]", expected "Union[Set[int], Set[str]]")
And yes, I know this would be better solved with a generic type variable instead of the unions. It is just to provide an additional demonstration of what I consider a bug.
My description of the problem may be poorly worded. Please feel free to correct me, so that I can describe it more precisely.
This issue seems vaguely related: #6463
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Reproduce the examples with mypy --strict, including the set/list calls and the custom generic function using Iterable and TypeVar. Trace the generic inference and union handling entry points; done means reveal_type preserves the union of int and str collections instead of widening the type argument to object.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- devtools
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100