RE: [wsrf] Issue: Finding WSDL from an EPR

["Savas Parastatidis" <Savas.Parastatidis@newcastle.ac.uk>]

Hi Igor,

> 
> Let us not mix the issues. WS-MetadataExchange or whatever else like
that
> is for run-time discovery by sending a SOAP message to (likely) the
> servive itself. That is fine, but has nothing to do with having a
place
> for a WSDL URL in an EPR.
> 
> For example, if I used a "registry" of sorts to discover an EPR and it
may
> already have a URL to the WSDL itself, why does it not just give it to
me?
> Why do I need to send a yet another SOAP message?
> 
> 

I don't understand your example here. With WS-MetadataExchange you don't
send a message to get a URL from which you can get the WSDL. A message
to the endpoint of the service (you have that in your EPR) will return
you the WSDL document. No need to send a message unless you need the
actual WSDL. But that's true with any approach.

If you put the EPR in a registry and you want to retrieve the WSDL of
the service that provides access to the referenced resource (*) then you
need to send a message to retrieve it. There are a number of ways of
doing this: (i) with WS-MD you can use the URI of the WSDL (e.g., use
HTTP GET if it is an HTTP URI), (ii) use WS-MetadataExchange and use a
SOAP message to retrieve the WSDL document (transport independence),
(iii) define a WS-RF specific way of doing it. Any way you choose, you
have to send a message to get the WSDL. So, I don't see how your
registry example identifies a problem.

Perhaps I misunderstood something.

Regards,
.savas.

(*) I noticed that many refer to "the WSDL of a resource" instead of
"the WSDL of a service providing access to a resource". Is this the
language used by WS-RF? I thought that the W3C defined WSDL to be a
language for describing the message exchange patterns with a service,
not a resource.