W3C home > Mailing lists > Public > xml-dist-app@w3.org > February 2003

Re: AFTF requirements, post-2003/02/07 telcon

From: Herve Ruellan <herve.ruellan@crf.canon.fr>
Date: Tue, 18 Feb 2003 09:49:57 +0100
Message-ID: <3E51F3B5.5020009@crf.canon.fr>
To: David Orchard <dorchard@bea.com>
CC: noah_mendelsohn@us.ibm.com, jones@research.att.com, xml-dist-app@w3.org


 From my understanding, there are two parts in R6.

First, each part attached to a SOAP message has to be identified by an 
absolute URI. That is, the 'name' of each part is an absolute URI. This 
is similar to the web, where each resource is identified by an absolute URI.

The second part of R6 deals about referencing or linking to those 
attached parts. R6 asks for being able to use any type of URI reference 
for making a reference or a link to a part. This includes:

- using an absolute URI reference, i.e. the idenfier of the part,

- using a relative URI reference, provided there is some means for 
building an absolute URI reference from it.

- using a URI reference containing a fragment identifier. In this case 
the target of the reference is only a subpart of the attached part.

Hope this helps,


David Orchard wrote:
> Sorry to drag this on.
> I'm still confused about what you mean.  On the web, resources are
> identified by URIs.  Not just by absolute URIs.
> I don't understand the point of requirement.  I had though that you meant to
> say "Any URIs used to refer to attachment resources must be absolute".  That
> is, it's a constraint upon strings that have type URI.  But you are also
> saying that you will allow relative resource identifiers in the later part
> of your message.
> Are you trying to say something like "URI strings of the full range of rfc
> 2396 are allowed in anything referring to attachment resources, so long as
> they can be absolutized.  That is, a relative URI is allowed if there is a
> base URI available for the dereferencing application to use".
> You know, I think we are violently agreeing.  I think you want to make sure
> that the URIs can be absolutized - not that the actual uri string is always
> in absolute URI format - and I'm saying that I want the full range of URIs
> available to people creating URI strings.  Which I think are almost the same
> thing, just worded differently.
> Again, it would be really nice if we had a term for a Absolute URI with or
> without a fragment identifier - maybe Absolute URI-Reference - then we could
> say All URI References must be dereferencable as an Absolute URI-Reference.
> Maybe we should feed this feedback into the editor of rfc2396 (Roy).
> Cheers,
> Dave
Received on Tuesday, 18 February 2003 03:50:52 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 22:01:22 UTC