Additional comment on LC76a

I noticed LC76a on tomorrow's agenda.  I like the general direction for
the way this issue is being resolved, but I think the wording needs to
be tightened up a little more, in addition to the tightening that
Jonathan Marsh already suggested[2].

Part 2 section 2.2 Fault Propagation Rules[1] says "extensions or
bindings may modify these rulesets", and later "Bindings, features, or
extension specifications may override the semantics of a fault
propagation ruleset".

Perhaps these statements should be say "mandatory extensions" and
"mandatory features" instead of just "extensions" and "features", since
optional extensions and features are supposed to be safely ignorable,
and extensions or features that cause faults to be sent to a different
place do not sound safely ignorable to me.  Or better yet, "mandatory
extensions and optional extensions that have been engaged" (and similar
for features).  

Basically, we don't want to imply that an optional extension would
change the semantics just by its presence in the WSDL document.  It
would have to be engaged by the client to do so.

(Hmm, now I'm wondering whether we should have defined a notion of
"engaged extension", which would cover both mandatory extensions and
optional extensions that have been engaged electively.  Should we?)


1.
http://dev.w3.org/cvsweb/~checkout~/2002/ws/desc/wsdl20/wsdl20-adjuncts.html?content-type=text/html;%20charset=utf-8#fault-rules
2.
http://lists.w3.org/Archives/Public/public-ws-desc-comments/2005May/0091.html


-- 

David Booth, Ph.D.
HP Software / Boston
Hewlett-Packard, Inc.

Received on Thursday, 9 June 2005 06:03:44 UTC