[fitsmime] Thoughts on FITS MIME types.

[William Pence]

Steve Allen wrote:
> 
> I am hoping to see some input from the VO community here, but at
> present my impression remains that the RFC should register
>         image/fits
>         application/fits
> and that the rest of the issues raised here are best solved by
> starting up classification efforts within the FITS community.

To add my $0.02 to this discussion, I would agree with Steve.  Also, I tend
to think image/fits should only be applied to 2-dimensional images,  since
there is a big step up in complexity (possibly requiring user interactions),
to deal with display of 3-D data cubes.

Regarding Random Groups, I currently don't see a strong need for a special
application/fits-group mimetype for that type of file.  I would think that
any application that could be associated with 'application/fits' should be
expected to not choke on a random groups FITS files.  The application would
not necessarily have to do anything intelligent with the data records, but
it should be able to at least deal with the primary header keywords, and be
able to move to any following extensions correctly.  This is all that fv is
able to do, and it is not hard to implement.  Basically, the application
just has to look for 2 keywords in the primary header:

NAXIS1 = 0
GROUPS = T

If both these are present, then the application should procede as if NAXIS1
= 1.  This then allows it to correctly compute the size of the random group
records, and move to any following extensions.  (With hindsight, it seems to
me it would have been much better to simply use the presence of the keyword

GROUPS = T

to denote the presence of random groups, instead of inventing the awkward
NAXIS1 = 0 business, but we're stuck with it now).

Bill
-- 
____________________________________________________________________
Dr. William Pence                          William.D.Pence at nasa.gov
NASA/GSFC Code 662         HEASARC         +1-301-286-4599 (voice)     
Greenbelt MD 20771                         +1-301-286-1684 (fax)