RE: Feature Composition Edge Cases

Thank you for the comment below, and for your patience with us in
resolving it.  We tracked the comment below as Issue LC20 [1].  The WG
agreed to change the feature composition model so that required
properties trump non-required properties, instead of the previous
proximity rules.  The editors have addressed the issue in their latest
drafts [2].

If you agree with our disposition of your comment, we'd like you to
acknowledge it within two weeks; otherwise we will assume you are
satisfied.  The WG plans to enter a second (short) Last Call period in
the near future, and we invite you to review that publication as well.


> -----Original Message-----
> From: [mailto:public-ws-desc-
>] On Behalf Of Asir Vedamuthu
> Sent: Monday, August 23, 2004 7:32 PM
> To: ''
> Subject: Feature Composition Edge Cases
> ref:
> 20040803/#Feature_composition_model
> Does our feature composition model capture all possible cases? I am
> not
> sure. Here is a sample edge case,
> <interface name="Bank">
>     <!-- All implementations of this interface must be secure -->
>     <feature uri=""
>              required="true"/>
>     ..
> </interface>
> <interface name="OpenBank" extends="Bank">
>     <!-- we don't give a damn -->
>     <feature uri=""
>              required="false"/>
>     ..
> </interface>
> According to Interface Component,
> "{features}	= The set of Feature components corresponding to the
> feature
> element information items in [children], if any, plus the set of
> Feature
> components in the {features} property of the Interface components in
> {extended interfaces}, if any."
> According to our equivalence rules, feature declared in Bank interface
> is
> not equivalent to the feature declared in Open Bank interface.
> Because, the
> value of {required} property is different. If these two feature
> components
> are present in interface component.{features}, what is the net effect?
> Secure channel is an optional feature! That confuses me. Please
> revisit our
> feature composition model and flush out all such edge cases.
> Also, shall we provide a special rule for computing the equivalence of
> feature components?
> Regards,
> Asir S Vedamuthu
> asirv at webmethods dot com

Received on Tuesday, 3 May 2005 19:57:18 UTC