DatetimeIndex constructor should accept set

[amgcc]

Describe the bug
Providing a set to the DatetimeIndex constructor is rejected.

To Reproduce

1. Provide a minimal runnable pandas example that is not properly checked by the stubs.

import pandas as pd


def main() -> None:
    s = {pd.Timestamp(2020, 1, 1), pd.Timestamp(2021, 1, 2)}
    l = [pd.Timestamp(2020, 1, 1), pd.Timestamp(2021, 1, 2)]
    idx = pd.DatetimeIndex(s)
    print(idx)

    idx2 = pd.DatetimeIndex(l)
    print(idx2)


if __name__ == '__main__':
    main()

2. Indicate which type checker you are using (mypy or pyright).
   mypy
3. Show the error message received from that type checker while checking your example.

dtidx_ctor.py:7: error: Argument 1 to "DatetimeIndex" has incompatible type "set[Timestamp]"; expected "ExtensionArray | ndarray[Any, Any] | Index[Any] | Series[Any] | ndarray[Any, Any] | list[Any] | tuple[Any, ...]"  [arg-type]
Found 1 error in 1 file (checked 1 source file)

Please complete the following information:

• OS: [e.g. Windows, Linux, MacOS] Linux
• OS Version [e.g. 22] Fedora 7
• python version 3.11
• version of type checker 1.8
• version of installed pandas-stubs 2.1.4

[Dr-Irv]

There's a related open issue on pandas about allowing sets as arguments for indexes:
pandas-dev/pandas#55425

The problem here is that an Index is ordered, so providing a set doesn't allow the order to be determined. Here's an example with pd.Index().

>>> pd.Index(set([1,2]))
Index([1, 2], dtype='int64')
>>> pd.Index(set([2,1]))
Index([1, 2], dtype='int64')

We decided for the stubs to disallow sets because the behavior for pandas is ambiguous.

[amgcc]

We often use DatetimeIndex in our application for union operations between calendars. In this case, the ordering isn't important. I think there is nothing "wrong" about using a set to initialize an index. Please reconsider.

[Dr-Irv]

We often use DatetimeIndex in our application for union operations between calendars. In this case, the ordering isn't important. I think there is nothing "wrong" about using a set to initialize an index. Please reconsider.

We discussed this in the pandas dev meeting and the intent is to (eventually) disallow sets with Index constructors. The recommendation for your code is to listify the arguments to the constructor, e.g.,

import pandas as pd


def main() -> None:
    s = {pd.Timestamp(2020, 1, 1), pd.Timestamp(2021, 1, 2)}
    l = [pd.Timestamp(2020, 1, 1), pd.Timestamp(2021, 1, 2)]
    idx = pd.DatetimeIndex(list(s))
    print(idx)

    idx2 = pd.DatetimeIndex(l)
    print(idx2)