Re: Something else to consider

Jim Webber wrote:

>  Umit:
>  
>  The proposal allows the following:
>
> <attribute name="foo">
>     <get body="x:fooType"/>
>     <set body="x:barType"/>
> </attribute>
>  
> I think you might be right. But I think the spirit of what Savas and 
> Sanjiva have said is probably right too. How about a simple syntactic 
> resolution to Savas's suggestion:
>  
> <attribute name="ncname" access="get|set|both"
>     [(body="qname") | (element="qname")]>
>         [<xsd:complexType> ... </xsd:complexType>]
> </attribute > 
>  

Jim,

This looks like what I suggested [1]. I think putting the keyword as 
access is a good idea. It also implies whether a particular attribute is 
queriable or settable. :-)

> Thus you'd have attribute declarations like:
>  
> <attribute name="foo" access="set" body="f:FooType"/>
>  

I would like to explore answers to my questions further as posted in [1].

-- The exact content of the get message exchange
-- The set message exchange.
-- SOAP HTTP binding and name resolution (with respect to keywords.) We 
need to do this regardless of whether we satisfy the multi attribute 
query requirement from [2].


> Now the other issues about where WSDL stops and other specs start I 
> think is actually relatively clear cut. I think if WSDL can allow 
> third party specifications to achieve what they want to achieve (the 
> Grid community immediately springs to mind as a group that WSDL 
> can/will be able to support), then that's as far as it needs to go. At 
> the moment there is a strong push from some user communities (In fact 
> this mailing list is symptomatic of that push :-)) to refactor some of 
> their own "vertical" issues into the "horizontal" WSDL. That is not to 
> say that such feedback is invalid, but care should be taken by the WG 
> that it does not become the vehicle for the advancement of one 
> particular use-case at the expense of others.
>  
> Personally I believe that the extensibility model in WSDL is the way 
> that third parties should primarily embrace and extend WSDL. I would 
> be all ears if people talked about improving the extensibility model 
> since it would benefit all user communities. However some significant 
> elements of the feedback to the WG doesn't seem to be pitched at the 
> meta-level, and instead suggests changes to the nuts and bolts of WSDL 
> which others may rely on. Thus the scope of the WG may have to be, 
> necessarily, brutally defined to prevent feature creep from verticals.
>  

I think you may consider talking to the properties and features folks. 
Their focus is to allow extensibility in a more pluggable way.

The problem is one person's nuts and bolts is the other person's 
extension :-). What you may consider is a vertical is not necessarily a 
vertical, it just looks like this because one group may see the need for 
nuts and bolts, can not find it in WSDL and instead develops a spec for 
it elsewhere. This is especially problematic if several groups start 
building similar extensions, not in WSDL, but elsewhere. There are many 
specifications that extend WSDL today. I personally find it particularly 
problematic, especially for interoperability. When vendors have to build 
several layers of the web services stack, you immediately know the holes.

Therefore, I think WSDL will benefit from incorporating some of this 
type of work as the nuts and bolts, instead of leaving it to extensions.

Again, this is a philosophical conversation and I would prefer we first 
tackle the problem at hand, namely the message structure and content for 
attributes.

> Jim
>  

--umit
References:

[1]http://lists.w3.org/Archives/Public/public-ws-desc-state/2003Jul/0043.html
[2]http://lists.w3.org/Archives/Public/public-ws-desc-state/2003Jun/0061.html

>

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