RE: issue 168 proposal: xsi:type of external references in Encoding

Noah,

I now understand a little better what you are looking for.  I think what you
are looking for is a specification of how URIs that exist with
Package/Message and refer to something within the same
package/message/container could be then guaranteed to be dereferencable.
Does that summarize in different word the issue correctly?

I'm still against doing anything on this though.  My issue is that this
smacks of local/global reference problems.  We have at least 2 choices for
doing what you'd like: 1) Somehow annotate the reference with the scope; 2)
Pick a special scheme that is guaranteed.  SwA conveniently used cid: which
the MIME Packaging spec defined for it's purposes.

There's a scalability of references from document to package to ? to web.
We find a similar issue comes up in the use of XML inside J2EE archive
files, that we don't want to deref something on the web because it is
supposed to be in the package.  And it really should be in the package.

I don't want to deal with package scoped references in SOAP, I'd rather deal
with them in the packaging work.
As SwA uses MIME and a special MIME content type, it all works.  Should we
use DIME, then we'll probably need a new scheme for specifying package level
DIME references.  That's an interesting challenge for a specified length
packaging mechanism to figure out how to label and manifest the chunks.  I
expect that an XML in DIME proposal would have such language.

If you want a framework for expressing constraints on references in packaged
content, it's certainly a bigger task than XMLP.

And again, it's not in SOAP 1.2 imho.

Cheers,
Dave


> -----Original Message-----
> From: Noah Mendelsohn [mailto:noah_mendelsohn@us.ibm.com]
> Sent: Wednesday, December 12, 2001 9:25 AM
> To: dorchard@bea.com
> Cc: andrewl@microsoft.com; jacek@systinet.com; xml-dist-app@w3.org
> Subject: RE: issue 168 proposal: xsi:type of external references in
> Encoding
>
>
> David Orchard writes:
>
> >> I'm confused about what you would like XMLP to do.
>
> In the core protocol, nothing, except to indicate what feature
> specifications should do.
>
> >> I posit that XMLP cannot/should not define any
> >> kind of additional elements/attributes for
> >> purposes of encoding information about references
> >> and hints/rules on what to do with references.
>
> Agreed.
>
> >> SwA does not indicate how/when to indicate that
> >> references are to be dereferenced.  It talks
> >> about URI resolution, not dereferencing.
>
> Agreed.  I believe I am talking about resolution as well.  I
> may not have
> written carefully.
>
> >> To me, it says that a receiver application may
> >> choose to follow certain rules for URI resolution,
> >> and it's up to the receiver application to
> >> know which references they should be applied to.
>
> Well, the use of a given URI scheme says a great deal about
> the resolution
> mechanisms to be used.  I think we'd agree on that.  So,
> "when" is at the
> discretion of the application, "how" is controlled by the URI scheme.
> Right?
>
> ...
>
> Let me net out what I'm looking for, because I suspect it's a
> lot less
> than you think.  I'm looking for a useful notion of
> "message", as distinct
> from all the other resources out there on the web.  I
> completely buy the
> notion that resources carried with the message, whether in
> the envelope or
> in attachments, have URIs and are referenced per web archictecture.
>
> The one one thing I'm trying to add is that, for each attachment
> architecture (SwA, DIME), we document to the extent practical
> the URI's
> that will be used to refer to information >>carried with the
> message<<.
> Otherwise, I can claim I'm doing SOAP+Attachments, use
> href="http://....",
> and say "well, I thought it was an attachment, too bad you
> had to go out
> on the web to get it."  I want to be able to say "If you're
> using S+A,
> then a URI of the form cid:// is an attempt to reference information
> carried with the message...at a given node where the message has been
> completely received, retrieval of the resources referenced by
> such URI's
> SHOULD NOT fail due to lack of network connectivity, etc."
>
> In other words, I'm looking for a simple abstraction,
> presumbaly based on
> URI's, to distinguish (an attempt to reference) information
> that is truly
> an attachment from an attempt to reference all the other useful
> information on the web.  That's all I'm looking for.
>
> Thanks.
>
> Noah
>
>
> --------------------------------------------------------------
> ----------
> Noah Mendelsohn                                    Voice:
> 1-617-693-4036
> Lotus Development Corp.                            Fax: 1-617-693-8676
> One Rogers Street
> Cambridge, MA 02142
> --------------------------------------------------------------
> ----------
>
>
>
>

Received on Wednesday, 12 December 2001 14:03:28 UTC