List concat with different types causes error

When I run mypy on this code:

from typing import List, Any

def f(a: List[int], b: List[str]):
    return a+b

f([1,2,3], ["3","4","5"])

I get this error:
/home/georgeatmit/type1.py: note: In function "f":
/home/georgeatmit/type1.py:3: error: Unsupported operand types for + ("list" and List[str])

Is combining lists of different types supported at the moment?

I think it's fine to get an error for this -- it's more likely to be a programming mistake than intentional. Maybe you can use `cast()` to remove the warning if this is really what you want?

I agree that this looks like a programming mistake, I was wondering about the general rule of combining collections with different types in Python's type hinting syntax, and a bit confused with the error message, why List[int] shows up as "list" but List[str] is represented correctly. I'll do some investigating and file another issue if I find anything.

from typing import List, Any

def f(a: List[int], b: List[str]) -> List[Union[str, int]]:
    return a+b

f([1,2,3], ["3","4","5"])

should be totally valid.

_T = TypeVar('_T')
_U = TypeVar('_U')

def list_add(a: Iterable[_T], b: Iterable[_U]) -> Iterable[Union[_T, _U]]:
    items: Iterable[Iterable[Union[_T, _U]] = [a, b]
    return itertools.chain.from_iterable(items)

f: Callable[[List[int], List[str]], List[Union[str, int]] = list_add

should do this without cast.

I wish this weren't closed.

List[A] + List[B] should be inferred as List[Union[A, B]]. Sure this could often simply be a mistake, but ideally that would be caught downstream when the resulting list would not match an expected type.

a: List[int] = [1, 3]
b: List[str] = ['foo', 'bar']
c = a + b # this should work fine with the implied type of List[Union[int, str]]
for item in c:
    print(item + 5) # this is where I would expect the mypy error

In my use-case I want to make the trade-off the headaches of supporting a mixed collection to support a more intuitive interface for the users of my API. But now I have to explicitly type c or case a and b in order to support that - adding to the headache.

Fix flake8 and mypy errors

Make WIC Participation Data Available to Frontend (#44)

Ran into the same issue by trying to have the possibility of None in the list as I'm generating schema.

Converting to tuple did work as tuples do support mixed types and undefined lenght so you can have Tuple[Optional[str], ...] but if anything this feels more of a hack as imutability does not matter in that particular case.

It's great to assume error by default, but types should accept mixed cases if provided as the language does.

This should not be closed, or should be linked to more appropriate bug report.

I agree that this should be re-opened

Any updates? It's still very relevant...
Just faced the same situation, and I don't actually want to create workarounds every time...

The original example typechecks for me now, likely thanks to python/typeshed#8293. If you have a different problem, please open a new issue.

Navigation Menu

Repository navigation

List concat with different types causes error #720

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

List concat with different types causes error #720

Description

Activity

gvanrossum commented on Jul 8, 2015

noisecapella commented on Jul 9, 2015

graingert commented on Jul 25, 2017

graingert commented on Jul 25, 2017

fivepapertigers commented on Jun 9, 2020

JackLeo commented on Oct 11, 2021

Randl commented on Jun 16, 2022

miyukiko-su commented on Oct 5, 2023

JelleZijlstra commented on Oct 5, 2023

Metadata

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

Issue actions