RE: Requirements WSP02, WSP03 & WSP08 (was Re: Web Service Definition)

Thanks, Hugo.  
You've condensed WSP02 & WSP03 quite well.
More inlined comments follow.
 
> -----Original Message-----
> From: Hugo Haas [mailto:hugo@w3.org]
> Sent: Wednesday, February 27, 2002 12:12 PM
[snip]

> >   WSP02: A web service MUST provide standards-based programmatic
> >   interface with well defined input/output parameters.
> >   (For the sake of simplicity, RPC returns may be deemed output 
> >   parameters in web services.)
> >   
> >   WSP03: A web service MUST be sufficiently well formulated such
> >   such it can be unambiguously described using WSDL.  
> 
> I think that WSP02 and WSP03 are not clearly separated.
> 
> They both talk about clear description. I would separate them as:
> 
>   WSP02: A Web service MUST provide standards-based programmatic
>   interface.
Good!

>   WSP03: A Web service MUST be sufficiently well formulated such [that] it
>   can be unambiguously described using a service description language.
> 
> Two changes:
> - "with well defined input/output parameters" has been removed from
>   WSP02; I believe that WSP03 covers this.
> - "WSDL" is replaced by "a service description language"; I am still
>   unhappy about this but here is my train of thoughts: which
>   specification does the current WSP03 refer to? WSDL 1.0? WSDL 1.1?
>   The output of the Web Services Description Working Group? The
>   description technology that our group will identify for the Web
>   services framework? My modification implies that it is the latter.

On WSP03, my reasoning is that a WS may be described "multi-lingually."
However, there ought to be one language that's official by virtual of
its universality (in terms of functional usage).  Wanting to be as
specific as the need for generality would allow, I called out WSDL
because I believed its ubiquity to the WS space was analogous to what
English is to the world.  (I could be off there.)  So, on a scale
with specificity at one end with WSDL and generality at the other with
"service description language," I'd rather err on the safer side (ala
generality), thus I find your change agreeable.  Thx.
 
> [..]
> >   WSP08: A web service MAY advertise itself in as many public
> >   directories as appropriate, such as those operated by the
> >   UDDI consortium.  
> 
> We should be careful about centralization:
> 
>   The Web is by design and philosophy a decentralized system, and its
>   vulnerabilities lie wherever a central facility exists.
>   -- http://www.w3.org/DesignIssues/Architecture#schemes
> 
> I don't think that WSP08 is really necessary, or at least not
> formulated as such. What about:
> 
>    WSP08: Web services MUST be able to advertize themselves.

WSP08 is necessary because the property to be described
and to be discovered differentiates the Web Service computing
model from other contenders, such as EDI, CORBA, ...

I was cognizant of the other Description & Discovery (D&D) models
(such as WS-Inspection) besides UDDI, which I only used as an
exemplary reference in WSP08.  What you brought up has flagged
that WSP08 needs some re-work, which I agree.  However, the
terse replacement you're suggesting shuts out applications
that want to be advertised NOT by themselves, but by their
agents.  I don't think you mean to do that.

So, perhaps we the WG can agree on D&D being a property of WS.
Can't we?  If we can indeed, then let's figure out the verbiage.  

Cheers,

Joe Hui
Exodus, a Cable & Wireless service
=====================================

> Regards,
> 
> Hugo
> 
> -- 
> Hugo Haas - W3C
> mailto:hugo@w3.org - http://www.w3.org/People/Hugo/ - 
> tel:+1-617-452-2092
> 
>

Received on Wednesday, 27 February 2002 17:39:18 UTC