Clive Page wrote:
> Martin Hill wrote:
> > However let us please please not make it the long term defacto
> > standard for astronomical data!
>
> I think that whether it becomes a de facto standard or not depends not
> on us but on the whole community.
Agreed, unfortunately.
> The FITS standard started out as purely a tranport format, but gradually
> became a viable storage and processing format as tools were developed...
*shake head*
Yes, FITS became a storage and processing format, however it was (and still isn't) particularly good at those jobs.
Unfortunately the community has made it a standard and we're stuck with it, one of the things I always list when I'm doing a "What is the VO going to do for us..." talk (depending on the audience) is "Get rid of FITS as a storage and processing format".
> > The big advantage of VOTable over FITS is a well-defined metadata area
> > and human-readability; the former necessary for the VO, the latter
> > useful for debugging but let's face it, is mostly useful because of
> > the lack of VOTable tools.
>
> I disagree with that. I think FITS metadata are just about as
> well-defined as those of VOTable, and if anything the headers of a FITS
> file are more legible...
The meta-data and (non-existant) data-model of a FITS file is totally horrible...
The format isn't self-describing, with the data model it represents was designed around the file format rather than the file format being designed around the data model (something unfortunately we may be replicating with VOTable), and the format itself is heavily overloaded.
I can't agree that FITS is a good file format, the only reason anyone gets anything done with it at all is that its a mature platform, with lots of tools available to hide the true horror from you...
> ...though that depends a bit on which tool you use to view them.
...which I guess this bears out.
FITS started out simply and has gotten hideously complex over time. If mature toolsets weren't around to handle it things would be very different. Anyone writing a FITS library today from scratch has an uphill job to to handle even the "most common" cases.
You only have to look at ESO HIERARCH headers to know how badly FITS has departed from its original intentions.
> ...and as a result some FITS-using communities have defined their own
> ad-hoc standards.
You say this like it's a good thing!?
Okay, this turned into a rant, but FITS is horrible. The last thing we want to do isemulate the FITS standards process.
Al.
-- Dr. A. Allan, School of Physics, University of ExeterReceived on 2004-04-14Z12:16:17