W3C home > Mailing lists > Public > xml-dist-app@w3.org > September 2001

Re: Encoding XML in URIs

From: Mark Nottingham <mnot@mnot.net>
Date: Wed, 5 Sep 2001 18:10:02 -0700
To: Paul Prescod <paulp@ActiveState.com>
Cc: XML Distributed Applications List <xml-dist-app@w3.org>
Message-ID: <20010905180957.C1444@mnot.net>

I don't know that there's a deep reasoning against this -- I'll leave
that to the URI experts -- but IMHO it's a bit tortured to
deconstruct an XML message into XPath components, encode them into a
URI, make a reference, and then reconstruct a DOM tree on the other
side, just to gain idempotency semantics.

On the surface, it certainly seems valid; the interesting part of
2396 is section 3.4, 'Query Component'.

 The query component is a string of information to be interpreted by
 the resource.

This is the only component which might be interpreted as using URIs
for data transmission; every other component is tasked for
identification purposes. I think there are reasonable bounds as to
the nature and amount of data that can reasonably be expected to be
transmitted by this mechanism. You're right in that it hasn't been
done before; looks like we're starting to test it now. 

I think some of the resistence is because we're currently designing
SOAP's *default* HTTP binding. This doesn't preclude other transport
bindings based upon HTTP. However, SOAP is a generic XML messaging
framework, and that means that it should accommodate a number of use
cases. If the default binding for SOAP were to be URI-encoded, it
would force SOAP applications into very specific uses (tied
intimately to HTTP semantics), rather than being a generic,
transport-independent framework. It would be a poor example as to
what SOAP can do.


Perhaps my call to send this to the TAG was somewhat premature; my
motivation was to assure that transmission of XML data in URIs is
compatible with the Web architecture, and if it is, that a single,
well-designed solution come about, since it does seem to be of
interest to all.

(and something in the back of my head wonders that if this were a
solution that embodies the Web architecture, why this WG isn't called
'URI Protocol' ;)



On Wed, Sep 05, 2001 at 05:12:29PM -0700, Paul Prescod wrote:
> Mark Nottingham wrote:
> > 
> > I am *very* reluctant to re-start the 'web infrastructure'
> > discussion, but XForms' work in URI encoding seems relevant;
> > 
> >   http://www.w3.org/TR/2001/WD-xforms-20010828/slice11.html#rpm-send-urlencoded
> > 
> > IMHO there are still fundamental and practical issues with this use
> > of URIs, and personally I think this is something the TAG should
> > address as soon as it gets instantiated, as it seems to be of
> > interest in more than one context.
> 
> I'm curious about whether anyone knows of a "deep" reason that URIs
> cannot merely have components that are deeply hierarchical and nested?
> Just because that has never been done before?
> -- 
> Take a recipe. Leave a recipe.  
> Python Cookbook!  http://www.ActiveState.com/pythoncookbook
> 

-- 
Mark Nottingham
http://www.mnot.net/
 
Received on Wednesday, 5 September 2001 21:10:04 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:59:03 GMT