----- Original Message -----
Sent: Wednesday, April 14, 2004 4:21
PM
Subject: RE: [ws-caf] issue 84
The
context question was more the concern of what is now issue 92 (PRF-34), though
it's only thrown in at the end of the explanation of that. CAF-84 was
more concerned with the definition of things, so
rathole-identifier : a URI that identifies where the rats
are
probably just means it is an unambiguous name for the rathole, but it's
possible that some other bit of the specification requires implementations to
dereference it (presumably to get the rats out). The definition of the
identifier should say whether the required use is just identification.
(the
distinction isn't necessarily rigid - as with xml namespace ids, where the
requirement is only identification, but if you choose to make it an http url
AND you choose to put a copy of the schema (or whatever) on the page pointed
to, you may be thought of as helpful)
Peter
I thought you were hinting at the problem that
we currently overload the context identifier field so that you can't tell
from:
<wsctx>
<context-identifier>foo</context-identifier>
</wsctx>
whether foo is a rereferencable URI or just a
basic context without any augmentation. Seems to me that is an issue and we
either add an attribute to context-identifier or have a different
element.
Mark.
----
Mark Little,
Chief Architect,
Transactions,
Arjuna Technologies Ltd.
www.arjuna.com
----- Original Message -----
Sent: Wednesday, April 14, 2004 2:46
PM
Subject: RE: [ws-caf] issue 84
When I raised the issue I was thinking primarily that it must be
stated in the definition of a field, rather than that it need be in the
schema as well. I've no objection as such to having it in the syntax, but
I'm not sure it's needed.
Peter
This is about making it explicit whether a
URI is for referencing or not. I agree it needs to be clarified and
favour an optional attribute like:
<wsctx>
<context-uri
reference=true/>
</wsctx>
I think the majority of uses for URI in the
raw context are for unique identification purposes, so it's probable
that this is only actually required on the context-id field. The
default IMO would be false.
Again, I want to set up a Kavi ballot for
this in a day or so, to try to clear as many of the "minor" issues as
possible before the f2f.
Mark.