Re: Agenda, 5 June 2003 WS Desc telcon

Arthur Ryman wrote:

>
> Umit,
>
> I replied to your point about whether R085 is dynamic or static in [1] 
> which has a more accurate Subject.
>
> [1] http://lists.w3.org/Archives/Public/www-ws-desc/2003Jun/0148.html
>
> Arthur Ryman 

Arthur,

I disagree with your assesment. This is why I don't want to mislead the 
wg by replying to the message you posted to that thread :-).  See my 
comments below instead.

>
> Umit,
>
> The notes in the thread [1] are long and contain several issues, so 
> I'm starting a new thread to discuss the subject issue. Here is the 
> text of R085 taken from [2]: 

The name of the thread you have chosen is misleading. I also don't think 
that  threads based on ambiguous requirements would get us anywhere. 
Read on. 

>
>
> R085
> The description language SHOULD allow describing Messages that include 
> references (URIs) to typed referents, both values and Services. (From 
> PP. Last discussed 11 April, 2002.)
>
> I'll parse this.
>
> "The description language" == WSDL
>
> "SHOULD allow describing Messages that include references (URIs) to 
> typed referents," == a Web service message may contain a URI, in which 
> case WSDL should be able to describe the type of thing identified by 
> the URI
>
> "both values" == the URI could refer to a document, in which case we 
> should be able to describe its data type, e.g. MIME type or  XSD
>
> "and Services" == the URI could refer to a Web service, in which case 
> we should be able to describe to its WSDL interface and binding
>
> This requirement clearly states that the WSDL document should be able 
> to describe the type of a Service refered to by a URI. This is static 
> typing. 

 The requirement says that the language should allow references to typed 
referents. The only thing it says is that the URI must represent a value 
or a service. The requirement as the way it is written does not state 
that the type of the service should be inferred as well. It only says a 
reference must be identified as a service, not the type of the service. 
So this last bit is not inferrable from the definition.

Further, the subject of your thread [1] made one more inference that 
does not appear possible with this definition. R085 does not say 
anything about endpoint URIs. At best, R085 is talking about 
distinguishing between values and references. What if the reference is 
by using a URI which describes a service, not the URI of the endpoint? 
The description still applies. I think you made this inference because 
your proposal is to indicate the types and bindings for a URI which is 
in essence corresponds to an endpoint URI. However, the requirement does 
not indicate a solution.

>
>
> This requirement does NOT state that the Message should be 
> self-describing, i.e. that by looking at the message instance that 
> contains the URI we can also find the additional type information, 
> i.e. WSDL interface and binding. That would be dynamic typing. 

The requirement as it is written states at best that one should be able 
to treat service references differently than values. Typed URIs are a 
problem in general.

>
>
> However, I agree that the dynamic case is important and hence 
> WS-Addressing [3]. IMHO, it would make sense for the WG to address 
> that case. 

I am glad that we are in agreement there with respect to the importance 
of dynamic addressing. However, I am afraid that making the reference to 
WS-Addressing is not useful to me or to anyone else. This is not a WS 
specification, or a w3c specification. Further, the IP requirements may 
make it nonusable.  Just referencing to it does not allow anyone who 
wants to be able to use a web service reference. What can I do with it? 
At the moment, nothing. Maybe you can help with this as you have been 
sending a reference to the document. May I also request you not to send 
a reference to WS-Addressing again unless there is a clear way for rest 
of us to use it?

I propose that we have an issue with respect to R085. From this 
discussion, it is obvious to me that it is ambiguous and is subject to 
different interpretations as you have  illustrated. I would like be able 
to define web service references as a first class citizen in WSDL and 
use them uniformly, statically or dynamically. Having one solution for 
one but not the other is at best a half baked solution. Hence the 
requirement should reflect both cases.

Cheers,

--umit

[1] http://lists.w3.org/Archives/Public/www-ws-desc/2003Jun/0148.html


-- 
Umit Yalcinalp                                  
Consulting Member of Technical Staff
ORACLE
Phone: +1 650 607 6154                          
Email: umit.yalcinalp@oracle.com