On Tuesday 21 August 2007 19:10, Steve Allen wrote:
A question for those who have been around longer than I:

Does "once FITS always FITS" mean

I'm may not fully qualify but for me the point was archiving data
in FITS format. That is, if one at a certain moment writes a data
file in FITS format one should be able to read it N years after
with a standard reader without problems. This would come
close to
* Anything ever explicitly allowed can never be disallowed

It is almost impossible to write a standard which cannot be victim
of marginal interpretations (FITS being no exception). Thus, one
should not disallow something which have been in frequent use.
Bordercases where the standard has been stretched an
unreasonable amount must be considered individually. If it is
clearly against common interpretation a clarification (as in the
NOST documents) is necessary.

In conclusion, changes should be backwards compatible i.e.
not invalidate old files. This is my main concern with respect
to FITS versions and the like (e.g. using the data of writing
as a flag for what is allowed or not).

Preben