I don't know if this is a bug or is expected, but in astropy 3.0 this:
from astropy.io import fits
hdr = fits.header()
hdr.set('A')
print(hdr)
gets you this
A = ' '
END
but in 3.2 you get this
A =
END
with no value. Is this valid FITS?
EDIT: Fixed code
Hmm... I see that in 4.1dev too. Do you want to git bisect
and see which commit was the culprit?
Yes, this is valid FITS, it creates an undefined card : #8572
Yes, the commit is 545f488cbb4a5ccca908bffeb90bc29ad41bd9c1
I see that before this commit a value of None
in this function was mapped to ' '
It's my understanding that this is the intended behavior, so this can be closed.