Simon Conseil

Yes this kind of error is hard to test properly... but the risk seems limited, this code would not affect normal behavior and we can adjust in the future if...

Yes let's merge this PR and we can still reconsider in the future. And let's hope we won't have to support all those error cases :sweat_smile:

Not exactly the same, as it is usually not installed directly in the source directory, but it can be managed by pip/setuptools: https://wheel.readthedocs.io/en/stable/#including-the-license-in-the-generated-wheel-file Adding this would allow to have the...

Depends what the decorator is for and what problem we are trying to solve, which isn't clear to me right now. I guess one issue to solve would be to...

Serialization for what ? YAML is used for a few specific cases where I don't think it can be replaced by any random (binary) serialization format.

Imho it doesn't sound like a good idea to hide deprecation warning from users. Yes with PEP 565 they will sometimes see the warnings from an interactive session but not...

> I'm not sure that there are many cases where a user can so usefully act on a warning that it is needed just to take an example of changes...

> in general feel that one should follow the standard unless there are very good reasons not to do so, and AstropyDeprecationWarning just shouts "subclass of DeprecationWarning From Python docs,...

The FITS compression code is being heavily refactored in #15474, so let's ping @astrofrog :)

It looks like `TableHDU.from_columns(hdu.columns)` works, and `BinTableHDU.from_columns(hdu)` as well, so `TableHDU.from_columns(hdu) ` should also be valid I guess. So I think it's a real bug, not a docstring issue.