Subclassing DataArray for purely type-checking purposes

[maresb]

I wrote some code that looks roughly like:

class PreprocessedArray(xr.DataArray):
    """A DataArray which has gone through preprocessing"""

def preprocess(a: xr.DataArray) -> PreprocessedArray:
    result: xr.DataArray
    ...
    return cast(PreprocessedArray, result)

def process(a: PreprocessedArray):
    ...

Note that PreprocessedArray is never instantiated or used at runtime. This is purely for type checking so that I will be alerted by my IDE's type checker in case I try to run process on an array which hasn't yet been preprocessed.

Unfortunately I get the warning

FutureWarning: xarray subclass DA should explicitly define __slots__
  class PreprocessedArray(xr.DataArray):

This isn't so big of a deal, and I can just add the line

    __slots__ = ()  # Suppress slots warning

to my class definition, but it feels really out-of-place. Am I doing this right? I saw a bunch of discussion at #3980, and I was wondering if this use case has been considered. (Making a subclass which is never used.)

[maresb]

I'm fairly satisfied with this solution I came up with:

from typing import TYPE_CHECKING

if TYPE_CHECKING:
    class PreprocessedArray(xr.DataArray):
        """A DataArray which has gone through preprocessing"""
else:
    PreprocessedArray = xr.DataArray