"Arguments missing" for optional arguments in dataclass

[talarari]

Environment data

• Language Server version: Pylance language server 2021.1.3
• OS and version: Mac os catalina
• Python version (& distribution if applicable, e.g. Anaconda): 3.7.8

Expected behaviour

Creating an instance of a dataclass without arguments for optional fields will show no error

Actual behaviour

creating an instacnce of a dataclass without arguments for optional fields shows and error "arguments missing"

[notatallshaw]

Have you tried running this code?

When I run this:

from dataclasses import dataclass
from typing import Optional

@dataclass
class User:
    name: str
    age: Optional[int]

user = User(name="Jon")

I get this output:

Traceback (most recent call last):
  File "data_class_test.py", line 9, in <module>    
    user = User(name="Jon")
TypeError: __init__() missing 1 required positional argument: 'age'

Optional from typing means the value can by the given type (int in this case) or it can be the None type.

For a field to be actually optional in the dataclass constructor you must give it a default value, in this case probably None:

@dataclass
class User:
    name: str
    age: Optional[int] = None

Which Pylance correctly handles (actually if this was a bug I assume it would be with Pyright not Pylance).

[erictraut]

The notation Optional[int] means that the value must be of type None or int. It does not mean that it can be omitted from the dataclass. There is currently no way to specify that individual elements are not required. There is an ongoing discussion in the typing-sig about adding such a facility in Python 3.10, but no decision has been finalized yet.

Pylance is doing the right thing here based on PEP 557.

[erictraut]

If you're interested in following the discussion in the typing-sig, here's a link.

[talarari]

thanks for the info, using a default value makes sense.