R: Theoretical access protocol

[Miguel Cerviño]

Dear Patrizia

Yes, in fact, TSAP is used in the SVO for access to stellar  
isochrones (tracks should be
similar).


> I think is a good choice making the query in separated and linked  
> steps for
> any type of simulated data and add this capability to the VO tools  
> that can
> visualized or analyse these kind of data.

:) From my point of view, it is just know the problem :) that  
applications accept
these queries and not only the "cone search like" access to data.

> The problem could be that the
> output format of the simulated data could be very different, for  
> example
> Aladin is able to open only 2-D data, or catalog, so the steps of  
> the query
> must go to a 2-D file (if exist or there is some service able to  
> create it)
> like maps of N-D files like Gadget format or HDF5, or create only a  
> catalog
> of simulated objects.

I do not understand the point: the final TSAP result is an VOTable  
with a list of available data
(basically one of the fields in the resulting VOTables is just a link  
to somewhere to retrieve
the data file, usually in VOTable format)

In Aladin, you can load an isochrone table (using VizieR button), but  
Aladin is no able
to plot it, of course. But I usually do is load the isochrone VOtable  
in Aladin and, after that, I send it to TOPCAT using PLASTIC, since  
Aladin is the only VO access to VizieR...

Regarding the issue about the "native" format of the output, it is  
supposed that all data are
in VOTables ! It means, the services like Gadget or HDF5 must provide  
the data in VOTable, that is the metadata + the data itself (maybe in  
FITS, encoded data or a link to a file that
has any other format). Of course, it is up to the service developer  
how to write the
format with the data, the more compilant with the VO (mainly html- 
like or FITS), the more
applications will use the service ...
Summing up, there must be a conversion layer between any program  
output and its VO access.

May be I am wrong, but one of the most power of the VO (in my point  
of view) is just to
do not worry about make specific drivers to use data depending the  
code that had generated it, but only worry about a driver to convert  
VOTables to the format I typically use in my
analysis tools (if they are not VO compliant)

cheers

miguel