Thoughts on FITS Undefined card

Astropy understands FITS header cards with undefined values. It assigns a card with a special value of astropy.io.fits.card.Undefined. This is perfectly fine until you try to use the dict interface to an astropy.io.fits.header and don’t know where the dict comes from. The astro_metadata_translator interface is designed to work for any “header” that looks like a dict. This can mean Astropy headers, LSST PropertyList and AST FitsChan. I’d like to write code that is like:

if k in header and header[k] is not None:
    ...

but at the moment I have to special case this just for Astropy:

if k in header and header[k] is not None and not isinstance(header[k], Undefined):
    ...

Would it be bad if astropy.io.fits.header.__getitem__ returned None for an Undefined header and __setitem__ converted None to Undefined internally? This would make the dict interface look just like a normal Python thing and I don’t think it’s unreasonable for an undefined value in a FITS header to map to a Python None.

I can do a PR if there are no objections.

[timj]

Ok. I’ll work on a patch.

[timj]

I think you need to read the paragraph that follows the one you pasted:

The value field, when present, shall contain the ASCII-text representation of a literal string constant, a logical constant, or a numerical constant, in the format specified in Sect. 4.2. The value field may be a null field; i.e., it may consist entirely of spaces, in which case the value associated with the keyword is undefined.