[fitsbits] Proposed Changes to the FITS Standard

Doug said:

Regarding versioning - I don't think FITS has changed enough to
warrant versioning.

I tend to agree, but wanted to point out that such would be needed
for any new feature or restriction that seriously challenges "once
FITS, always FITS". Mark has done a good job of characterizing what
versioning might mean.

Steve Allen discusses versioning based on conventions or vocabularies:

I would like to see FITS files be sufficiently regimented that they
can be sucked into a data structure which resembles a normalized
relational database, but that notion simply was not there from the
start. Again, that sort of thing could be accommodated by adding a
registry of conventions and a way of asserting which conventions
are in use.

or

explicitly declaring that certain FITS files are subsets of the
FITS standard which conform to a specific vocabulary.

Another option would be subclassing FITS as something like
"Encapsulated FITS" - either relying on a commenting convention as
Postscript does (since PS is a programming language), or on keyword
tags (since FITS is an explicit data structure).

Rob