<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta http-equiv="Content-Type" content="text/html;charset=ISO-8859-1">
<title></title>
</head>
<body text="#000000" bgcolor="#ffffff">
Hi Tara,<br>
<br>
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. <br>
<br>
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.<br>
<br>
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.<br>
<br>
I am interested in hearing your an other people's thoughts.<br>
<br>
Hans-Martin<br>
<br>
<br>
Tara Murphy wrote:<br>
<blockquote type="cite"
cite="midPine.SOL.4.33.0405260705520.8898-100000@venice.tip.CSIRO.AU">
<pre wrap="">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
</pre>
</blockquote>
<br>
<div class="moz-signature">-- <br>
<title>Dr. Hans-Martin Adorf</title>
<font color="#999999" face="Helvetica, Arial, sans-serif"><small> Dr.
Hans-Martin Adorf • Max-Planck-Institut f. extraterrestrische Physik </small></font><font
color="#999999" face="Helvetica, Arial, sans-serif"><small> • </small></font><font
color="#999999" face="Helvetica, Arial, sans-serif"><small>Giessenbachstr.
1 • D-85741 Garching b. München
• Germany <br>
Phone: +49-89-30000-3352 • Fax: +49-89-30000-3404 • E-Mail:
<a class="moz-txt-link-abbreviated" href="mailto:adorf@mpe.mpg.de">adorf@mpe.mpg.de</a></small></font><font color="#999999"
face="Helvetica, Arial, sans-serif"><small> • WWW: <a class="moz-txt-link-freetext" href="http://www.g-vo.org">http://www.g-vo.org</a></small></font>
</div>
</body>
</html>