inference on generic types depends on method returning generic

[Dr-Irv]

Environment data

• Language Server version: 2021.8.2
• OS and version: Windows 10 21H1
• Python version (& distribution if applicable, e.g. Anaconda): Anaconda 3.8.5

Example

This example is from pandas, but I don't think it is a pandas typing issue:

import pandas as pd

df = pd.DataFrame([[1, 2], [4, 3]], columns=["a", "b"])

s1 = df.max(axis=1)

s2 = df.min(axis=1)

s3 = s1 + s2

Expected behaviour

No error

Actual behaviour

pylance reports:

Operator "+" not supported for types "Series[Dtype@max]" and "Series[Dtype@min]"
  Operator "+" not supported for types "Series[Dtype@max]" and "Series[Dtype@min]"

The issue seems to be that the method that determined the type (max or min in this case) is being used to do the matching for the + operator. The operators only get matched if the types are the same (e.g., both max, both min, both __getitem__).

This could be a pandas stub problem, but it isn't clear how to fix it.

Logs

N/A

Code Snippet / Additional information

See above.

[Dzeri96]

I can confirm this issue on functions such as cumprod, cumsum etc.
Another strange error seems to be Operator "/" not supported for types "DataFrame" and "Series[Dtype@__getitem__]" when trying something like 1 / df.column.iloc[0]

[Dr-Irv]

@Dzeri96 can you open a new issue with respect to cumprod, cumsum, and / with examples that illustrate what you are seeing?

[Dr-Irv]

The issue above is now no longer occurring, but the type of s3 is Series[Timestamp], which is incorrect. So that's a different issue to fix.

[gramster]

Fixed in #196