Why not letget_tightbox return None instead working around the non-finite bbox?

@Tillsten you meanget_tightbbox of the axes children? I think thats a good idea as well, but I wasn't sure I'd be able to squash all possibleget_tightbbox implementations.

I think I'll do that for the baseArtist.get_tightbbox, but maybe keep this check too in case someone has done something special.

@Tillsten added the check in both places now. The one inAxes._get_tightbbox can't hurt (except maybe the eyes for the long if-statement).

Yeah I’m flexible about that. The original PR didn’t have this - it just checked for finite bbox widths and heights.

OTOH it’s not necessarily a bad idea to invalidate things we know we can’t do anything with. A NaN bbox width is not going to mean anything and pushing all the checks to a higher level just means more checks at a higher level.

A NaN bbox width is not going to mean anything and pushing all the checks to a higher level just means more checks at a higher level.

You'll have a newbbox is None check to do anyway at the higher level. If desired you could alternatively add anis_finite() method and checkbbox.is_finite(). (Or maybeis_valid() - Qt uses this naming convention.)

Edit: Would need careful consideration on the actual naming of such a method.

Right but we already check for None. Not sure why, though.

Removedartist.get_tightbbox returningNone if non-finite bbox.