W3C home > Mailing lists > Public > www-ws-desc@w3.org > December 2003

Re: Question on action item

From: Amelia A Lewis <alewis@tibco.com>
Date: Tue, 09 Dec 2003 10:16:05 -0500
To: Jacek Kopecky <jacek.kopecky@systinet.com>
Cc: www-ws-desc@w3.org
Message-id: <20031209101605.23eabcc1.alewis@tibco.com>

Dear Jacek,

On Tue, 09 Dec 2003 14:23:47 +0100
Jacek Kopecky <jacek.kopecky@systinet.com> wrote:
> I was wondering about the case where a service providing an operation
> that may result in faults is configured so that no faults are sent
> (presumably for security reasons). I don't think that the WSDL of the
> service should change because of this policy.

I understand this, but I don't think that this *can* or *should* be
expressed in the fault ruleset.  The fault ruleset, IMO, ought to be
unequivocal about the behavior expected of a service advertising a
particular MEP.  I could see a security feature redefining that
behavior.  But I can't see an "invisible" (not-advertised) feature
allowing the service to discard faults.  The client of the service has a
reasonable expectation of consistent behavior, based on advertised
(included-in-WSDL) description.  For an operation defined using
message-triggers-fault or fault-replaces-message, that expectation is
that when a fault is generated, it is sent, unless there is no path to
send it by.  If that behavior is advertised-as-changed by a required
feature (security-through-/dev/null), I could see it, but that's layered
on top of the ruleset, not built into it.

All IMO, of course.  Was this part of the discussion when the
'editorial' action item was created?  I don't think that it's editorial
....

Amy!
-- 
Amelia A. Lewis
Architect, TIBCO/Extensibility, Inc.
alewis@tibco.com
Received on Tuesday, 9 December 2003 10:20:20 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:58:27 GMT