Adding pandas Int64 support [fix #9280] - #9541

Thanks!

Test failure seems real, but change log can wait until the relevant section is available.

I'll let subpackage maintainer decides if a new test is necessary.

@bmorris3 - my sincere apologies for the late reply. This slipped off my radar and I just noticed it again today.

This looks like a great direction and we should go forward on it. Some issues that need to be addressed are:

• The current failing test. The current behavior (in master) is that an int MaskedColumn will round-trip through pandas and come back as a masked float. So what appears to be happening in your branch is that it comes back as an Object 'O' dtype. The desired behavior would be perfect round trip where it comes back as an int MaskedColumn.
• There should be a check on the pandas version to ensure that the installed version is >= 0.24 so that the Int64 type for Series is supported. You'll need to take a little care in testing with regards to the Pandas version since I don't think we have a strict requirements on pandas version for testing.
• A slight sticky wicket with regards to back-compatibility. We likely need a new keyword argument in to_pandas like use_nullable_int, which would default to None implying use it if possible (pandas version >= 0.24). False would mean never use it and True would mean always use it.

Hi humans 👋 - this pull request hasn't had any new commits for approximately 5 months. I plan to close this in a month if the pull request doesn't have any new commits by then.

In lieu of a stalled pull request, please consider closing this and open an issue instead if a reminder is needed to revisit in the future. Maintainers may also choose to add keep-open label to keep this PR open but it is discouraged unless absolutely necessary.

If this PR still needs to be reviewed, as an author, you can rebase it to reset the clock.

If you believe I commented on this pull request incorrectly, please report this here.

@bmorris3 - would you have time to rebase this and implement @taldcroft's comments to get this working?

@astrofrog @bmorris3 - I took a quick crack at fixing this. It looks like from_pandas() needs work as well as it is round-tripping back to astropy as an object column. Will see if this is an easy fix or not.

Passing tests locally! @astrofrog please have a look and think about what this might break.

@astrofrog -this needs a quick review, or we can just push to 4.2.

Thanks for carrying this over the finish line, @taldcroft!