RE: [xml-dev] dynamically generated XML Schema?! Re: [xml-dev] R: [xml-dev] Number of active public XML schemas

["Chiusano Joseph" <chiusano_joseph@bah.com>]

[Please see comment at end]

> -----Original Message-----
> From: Burak Emir [mailto:Burak.Emir@epfl.ch] 
> Sent: Thursday, November 04, 2004 3:15 AM
> To: Chiusano Joseph
> Cc: XML Developers List
> Subject: Re: [xml-dev] dynamically generated XML Schema?! Re: 
> [xml-dev] R: [xml-dev] Number of active public XML schemas
> 
> Joseph,
> 
> I agree WSDL documents should be generated rather than 
> written by hand, but doesn't schema provide means to pass 
> trees around whose structure you don't know (or at least, do 
> not need to know for the sake of web service invocation?).
> 
> Generating WSDL dynamically (like creating a new type of service at
> runtime) seems rather odd to me, because sure the invoker 
> program has to make sense out of the web service description. 
> If some form of negotiation takes place that leads to a new 
> service, the form of this negotiation is surely 
> predetermined, and you don't really need to create a new 
> schema for that.
> 
> It might make sense for contexts, ports (e.g. suppose google 
> rents out a searching web service with 20.000 instead of 
> 1.000 queries a day, they generate the WSDL after payment has 
> been confirmed.) But dynamically changing the type of the 
> data that goes over the wire does not happen there.
> 
> Maybe the point is that some specialization occurs (I am 
> inspired by what Mike Kay suggested, that later in a process 
> you might want to apply stricter validation). Suppose two 
> components talking to each other, finding out that they can 
> both deal with more specific messages, and then switch to the 
> new protocol. But again, this requires that both know the 
> protocol (the schema tht is more specific) in advance.
> 
> Maybe some unanticipated reconfiguration mechanism might 
> benefit from dynamically generated schemas, but I am still 
> missing a good example for such a reconfigurable component.
> 
> What would be an ad-hoc interaction that involves the change 
> in the type of messages?

One example that comes to mind is what I will call "message path
context" - that is, if a message from system A to system B was passed to
system A by system C, then an additional piece of information is
required to be sent to system B. That is, the path would look as
follows:

Without the context discussed here:
A -> B

With the context discussed here:
C -> A -> B

In more abstract terms, if we think of these as people making verbal
requests to each other, perhaps person C has an additional request such
as "and I would like that done within 3 days" (sort of like a
high-priority flag). Of course, the issue here is that system B may or
may not be able to interpret/accommodate such a request. I believe this
gets into Natural Language Processing and perhaps Matchmaking, some of
the things the Semantic Web Services folks are working with now. Whether
this will *ever* be possible is, I believe, still up for much debate.

Kind Regards,
Joseph Chiusano
Booz Allen Hamilton
Strategy and Technology Consultants to the World
 
> Chiusano Joseph wrote:
> 
> >I have often thought about the concept of dynamically generated WSDL 
> >documents, for cases in which more ad-hoc interaction among systems 
> >needs to occur, perhaps driven by the context of a request. 
> But I think 
> >we're a ways off from that in terms of standards and 
> products - if it 
> >is indeed a useful concept.
> >
> >Kind Regards,
> >Joseph Chiusano
> >Booz Allen Hamilton
> >Strategy and Technology Consultants to the World
> > 
> >
> >  
> >
> cheers,
> Burak Emir
> 
> http://lamp.epfl.ch/~buraq
>