Useless lines affect type checking

[titorau]

Versions
Pylance: 2025.10.4
Pyright: 1.1.407

Use case
I want to to export my dataclass instances to CSV. Hence, I write with the following code snippet in VSCode.

# Animals 1

import csv
from dataclasses import asdict, dataclass, fields
import sys


@dataclass
class Animal:
    name: str
    age: int

animals = [
    Animal(name="Foo", age=3),
    Animal(name="Bar", age=5),
]

field_names = [field.name for field in fields(Animal)]
writer = csv.DictWriter(sys.stdout, fieldnames=field_names)
writer.writeheader()
writer.writerows(
    map(asdict, animals),
)

Python runs it as expected and all function signatures are correctly detected by Pylance.

Expected behavior
There should be no error.

Current behavior
I get 3 errors at line 22, on map(asdict, animals).

Argument of type "Overload[(obj: DataclassInstance) -> dict[str, Any], (obj: DataclassInstance, *, dict_factory: (list[tuple[str, Any]]) -> _T@asdict) -> _T@asdict]" cannot be assigned to parameter "func" of type "(_T1@__new__) -> _S@map" in function "__new__"
  No overloaded function matches type "(Animal) -> _S@map"  (reportArgumentType)
Argument of type "Overload[(obj: DataclassInstance) -> dict[str, Any], (obj: DataclassInstance, *, dict_factory: (list[tuple[str, Any]]) -> _T@asdict) -> _T@asdict]" cannot be assigned to parameter "func" of type "(_T1@__new__) -> _S@map" in function "__new__"
  No overloaded function matches type "(_T1@__new__) -> _S@map"  (reportArgumentType)
Argument of type "map[Unknown]" cannot be assigned to parameter "rowdicts" of type "Iterable[Mapping[str, Any]]" in function "writerows"
  "map[Unknown]" is incompatible with protocol "Iterable[Mapping[str, Any]]"
    "__iter__" is an incompatible type
      Type "() -> map[Unknown]" is not assignable to type "() -> Iterator[_T_co@Iterable]"
        Function return type "map[Unknown]" is incompatible with type "Iterator[_T_co@Iterable]"
          "map[Unknown]" is incompatible with protocol "Iterator[_T_co@Iterable]"  (reportArgumentType)

Bug analysis
I have tested the 6 following variants which are basically doing the same things. Pylance and Pyright behave exactly the same, I only get errors in versions 1 and 4. The example I give uses csv and dataclasses but the issue seems much deeper since useless lines affect type checking.

• Animals 1
• Animals 2
• Animals 3
• Animals 4
• Animals 5
• Animals 6