W3C home > Mailing lists > Public > www-ws-desc@w3.org > February 2004

Re: Comments on WebArch

From: Amelia A Lewis <alewis@tibco.com>
Date: Tue, 24 Feb 2004 12:43:49 -0500
To: Jonathan Marsh <jmarsh@microsoft.com>
Cc: jean-jacques.moreau@crf.canon.fr, www-ws-desc@w3.org
Message-id: <20040224124349.56710aa2.alewis@tibco.com>

+1, with a reservation ... (see below)

On Tue, 24 Feb 2004 09:38:45 -0800
Jonathan Marsh <jmarsh@microsoft.com> wrote:

> I completely agree with our current desired behavior, just wanted to
> flag that there might be a boundary with what the AWWW recommends.
> There are analogous situations - for instance, I can open an XML
> document, stream it in, successfully process the information I find
> early in the file, and not report that 1Mb down in the file there is a
> well-formedness error.  The document is not well-formed, but a
> particular processor may not be able to detect and report the error.

I suppose that it's remotely possible that a WSDL would be ill-formed,
and still be processable, but well-formedness is the place where I tend
to want to make a firm stand.  I don't think that we should explicitly
permit ill-formedness, by any means.  I *do* think that we can permit
WSDL-validity errors (violations of constraints imposed by the WSDL
component model) and maybe even validity errors (violations of
constraints imposed by the schema for WSDL), if they are out-of-scope
for the processor, to be ignored.


> We're taking that too a new level by contemplating describing
> precisely what paths a WSDL processor might take and which errors it
> might not be able to detect.  Perhaps this detailed description of
> WSDL processing is the wrong approach, and we just need a generic
> statement that "a particular processor may not be able to detect and
> report errors in a portion of the document that it is not processing,"
> or "a particular processor is not required to detect and report errors
> in portions of the document that it is not processing."  And leave the
> definition of what parts of the document a processor processes to the
> processor.  This gets us out of the business of defining and
> constraining a WSDL processor, as DBooth is trying to do with issue
> 79.
> In any case, this might be worth a highlighting in our joint meeting
> with the TAG.
> > -----Original Message-----
> > From: Jean-Jacques Moreau [mailto:jean-jacques.moreau@crf.canon.fr]
> > Sent: Tuesday, February 24, 2004 6:02 AM
> > To: Jonathan Marsh
> > Cc: www-ws-desc@w3.org
> > Subject: Re: Comments on WebArch
> > 
> > At the same time, I wouldn't want, for example, to flag (potential)
> > errors in bindings that I not support. I would like to focus on the
> > binding I do support, and forget the rest, including the errors.
> > 
> > JJ.
> > 
> > Jonathan Marsh snip:
> > 
> > > 1.2.3: "Principle: Error recovery.  Silent recovery from error is
> > > harmful."
> > >
> > > Could this conflict with the ability to only use a part of a
> document
> > > that does not contain errors, without flagging errors in other
> > > parts
> of
> > > the document?

Amelia A. Lewis
Architect/Principal Engineer
TIBCO/Extensibility, Inc.
Received on Tuesday, 24 February 2004 12:43:32 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:54:47 UTC