more on FITS and VO

[Jonathan McDowell]

Mike wrote:

> 
> What if the spetrum were the primary data and I wanted the extraction region in the extension?

Tough - if you are describing it as image/fits, then all you
are guaranteed is the primary data.  If you want the extension information,
then use a service which returns it as application/fits. I may
not care about the extraction region, so I may be happy to use an
image/fits service to get it.


> there is no primary array in an NOAO Mosaic MEF
OK. So NOAO Mosaic MEFs are not a good thing to serve up as
image/fits, because you won't see any of the actual data that way.
They should always be encoded as application/fits (if we adopt
the present proposal). Adding a 1-pixel silly primary array 
would only make it valid-but-really-useless image/fits, it wouldn't
provide any useful info to someone using a minimally compliant
image/fits viewer.

    Jonathan