RE: Some requirements

I agree with Steve. In order for attribute definition in WSDL to be of any
use, people must be able to know, when they want an attribute, what message
to send to access them. Right? How do you define what a message looks like
in the web services world? With a piece of WSDL. It would be pretty ironic
for the WSDL spec itself to have English text that says "to retrieve an
attribute send a SOAP message containing as the first child of the body an
element called "attribute-name" the value of which is the name of the
attribute you want to retrieve" instead of just providing a WSDL operation
describing just that. Once this is done, people won't even necessarily have
to include this specific operation in their WSDL, we can state in the spec
that by indicating in their WSDL that they have attributes they implicitly
indicate their support of the WSDL operations needed to access attributes.

On the other hand, these operations should be limited to generic support for
any attribute, no specialized operation for "favorite" attributes. This
might be what Savas is warning us against?

Regards,

William

> -----Original Message-----
> From: Steve Graham [mailto:sggraham@us.ibm.com] 
> Sent: Thursday, June 12, 2003 4:49 PM
> To: Savas Parastatidis
> Cc: David Snelling; Jim Webber; Paul Watson; 
> public-ws-desc-state@w3c.org; 
> public-ws-desc-state-request@w3.org; Steve Tuecke
> Subject: RE: Some requirements
> 
> 
> 
> 
> 
> 
> 
> If we left the operatoins to access attributes up to myriad 
> application
> domains, there would be no consistency and therefore poor 
> interoperability.
> 
> Think of the analogy to SQL.  If SQL defined Data Definition 
> Language but
> left the Query language to be application specific, data 
> access would be a
> mess.
> 
> sgg
> 
> ++++++++
> Steve Graham
> sggraham@us.ibm.com
> (919)254-0615 (T/L 444)
> STSM, On Demand Architecture
> ++++++++
> 
> 
> 
>                                                               
>                                                               
>                    
>                       "Savas Parastatidis"                    
>                                                               
>                    
>                       <Savas.Parastatidis@newca        To:    
>    Steve Graham/Raleigh/IBM@IBMUS                             
>                    
>                       stle.ac.uk>                      cc:    
>    "David Snelling" <d.snelling@fle.fujitsu.com>, "Jim 
> Webber"                   
>                       Sent by:                          
> <jim.webber@arjuna.com>, "Paul Watson" 
> <Paul.Watson@newcastle.ac.uk>,                  
>                       public-ws-desc-state-requ         
> <public-ws-desc-state@w3c.org>, 
> <public-ws-desc-state-request@w3.org>, "Steve Tuecke"  
>                       est@w3.org                        
> <tuecke@mcs.anl.gov>                                          
>                          
>                                                        
> Subject:  RE: Some requirements                               
>                           
>                                                               
>                                                               
>                    
>                       06/12/2003 07:26 PM                     
>                                                               
>                    
>                                                               
>                                                               
>                    
>                                                               
>                                                               
>                    
> 
> 
> 
> 
> 
> >
> > Why do you feel this is domain specific?  Use of attributes cuts
> across
> > many domains.  Is this any more domain specific than CORBA IDL
> Attributes?
> >
> 
> I am sorry. I didn't make myself clear. I didn't suggest that 
> attributes
> are domain specific but, rather, that the definition of an interface
> with specific operations is, something that you suggested in your
> previous message.
> 
> WSDL should not include an interface with operations that services can
> choose to implement. That's not what WSDL does. WSDL is used 
> to describe
> interfaces and not to impose one.
> 
> I see attributes as a useful addition to WSDL but not all 
> features found
> in OGSI's SDE are easily transferable. I just can't see how operations
> that provide access to attributes can be specified by the WSDL
> specification.
> 
> A WS-Management or WS-Introspection specification, for example, could
> specify operations for accessing, quering attributes. These are
> application specific.
> 
> .savas.
> 
> 
> 
> 
>

Received on Thursday, 12 June 2003 20:36:54 UTC