Spectrum data model

[Mark Taylor]

On Wed, 13 Sep 2006, Matthew Graham wrote:

> Mark Taylor wrote:
> > On Wed, 13 Sep 2006, Matthew Graham wrote:
> >
> >   
> >> Hi,
> >>     
> >>> Otherwise, it seems to me that what is called for is an additional
> >>> field in the SSAP response which names the specific serialization 
> >>> format, if known.  This would require assigning some sort of name
> >>> to the XML, FITS and VOTable formats defined in the Spectral DM
> >>> document (presumably a URI of some sort).
> >>>
> >>> Mark
> >>>       
> >> The VOSpace specification uses uris to identify data formats and the 
> >> idea is that these will be registered somewhere, e.g. 
> >> ivo://net.ivoa.vospace/views/fits-table. It would make sense for both 
> >> groups at least to use the same uris.
> >>     
> >
> > Matthew
> >
> > Dave Morris made this point to me at the VOTech meeting in Strasbourg
> > last week, but I'm not sure I was thinking very clearly about it at
> > the time.  A question which occurs to me now: does VOSpace require 
> > a single data format per file (or entry, or whatever the correct 
> > term is), or can a single file be labelled with multiple different 
> > data format URIs at the same time?
> >
> > Mark
> >
> >   
> Hi,
> 
> If the data object is stored in a StructuredDataNode then multiple data 
> views are allowed: this is, of course, implementation dependent but the 
> idea is that if a VOSpace supported this, a FITS image might be 
> retrievable as a JPEG or a GIF, etc. or a FITS table as a VOTABLE or CSV.
> 
> If the data object is stored in an UnstructuredDataNode then essentially 
> bits in = bits out so only one view is allowed and that is the input view.

I wasn't thinking about data views (in the VOSpace sense) here.  
What I had in mind was the fact that assigning a single registered 
URI to describe the data type of a file in VOSpace may be problematic 
if the resolution of these types is sufficient to do things like
distinguish different FITS-based spectrum serializations.

The kind of data format identifier one would need to label a specific 
spectrum serialization, which is what I was suggesting in my original 
message, would have to be something with at least a level of detail like
"VOTable serialization of spectrum as defined by the Spectral DM standard",
and possibly be labelled by a version of that document as well.
This may be too specific for the kind of data type you'd want to
attach to a file in VOSpace, which I'd have thought would be nearer 
to MIME types, e.g.  "this is a VOTable".  Of course, it all depends 
on the requirements of the client that is retrieving that file, 
which makes it problematic to provide a single one-size-fits-all 
data type.

Mark

-- 
Mark Taylor   Astronomical Programmer   Physics, Bristol University, UK
m.b.taylor at bris.ac.uk +44-117-928-8776 http://www.star.bris.ac.uk/~mbt/