Web service standards question

[Hans-Martin Adorf]

Hi Tara,

we at GAVO have developed a simple Web service for accessing the ROSAT 
database. It implements a simple cone search. We are thinking of other 
Web services, e.g. a classifier. The main question in my mind is the 
format in which to pass in the information specifying the request and 
the format for the response. I personally do not think that VOTables in 
and out is the right answer.

VOTable is an XML format, and thus very appropriate for REST services, 
where the query is submitted via a simple HTTP GET or POST, and the 
result is commonly returned in XML. However, these days people, when 
talking about Web services, usually mean SOAP-based Web services, which 
I believe is what you are talking about, too. (Please correct me, if I'm 
wrong.) As the name states, SOAP is a protocol allowing the transfer of 
objects. I believe, we should not encode tabular data in a VOTable XML 
format, which is in some sense a serialization of some internal data 
object, and then let the server-side SOAP mechanism wrap it once more 
with XML. Not only is this totally inefficient (factors of 10 or more 
can easily be encountered), but we also do not take advantage of the 
abilities of the SOAP mechanism.

What I would like to see is an internal tabular data object that is 
populated on the server side, then transparently serialized and 
transmitted via SOAP, and finally reconstructed on the client side. 
There are examples for how such a tabular data object might look like, 
such as our GAVO table object, or the table objects used internally by 
the SAVOT code, or by the TOPTable code. If data arrives in a data 
object, no explicit XML-parsing is required. I would be interested in 
seeing such a common tabular data object adopted by service providers, 
which would simplify coding a lot. I can imagine that with a common 
tabular data object at hand it would be much easier to construct Web 
applications that bind together several existing services.

I am interested in hearing your an other people's thoughts.

Hans-Martin


Tara Murphy wrote:

>Hi,
>I have built a demonstrator web service to provide unit conversion
>functionality (from the AIPS++ Quanta module).
>This is part of the ATNF VO effort, and as the aim is to be VO-compatible,
>I was wondering what standards I should be following at this stage.
>
>The two areas I'm aware of are the StandardInterfaces document and
>VOTable. The service can take and return VOTables, so I think it is ok in
>that respect. Am I right in thinking the Standard Interfaces are still very
>much under development (and maybe not worth worrying about at this
>stage?)
>
>It would be good if somebody could summarise what else (if anything) I
>need to be thinking about.
>
>thanks,
>Tara
>
>-----
>Tara Murphy
>Postdoctoral Scientist
>
>Australia Telescope National Facility
>PO Box 76
>Epping NSW 1710
>Australia
>
>
>
>
>  
>

-- 
Dr. Hans-Martin Adorf . Max-Planck-Institut f. extraterrestrische Physik 
. Giessenbachstr. 1 . D-85741 Garching b. München . Germany
Phone: +49-89-30000-3352 . Fax: +49-89-30000-3404 . E-Mail: 
adorf at mpe.mpg.de . WWW: http://www.g-vo.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.ivoa.net/pipermail/grid/attachments/20040526/478fe83d/attachment-0001.html>