Re: proposal for restricting a service to a single interface

Arthur,

Thanks for the comments.

> Yes, the default value for @xpath is "." which selects the entire part.
That's in the proposal.
Is there a public link to the proposal ?
If XPath selectors are supported, then are multiple endpoints per single
(complex) part also supported ?

>Making the binding optional is much more problematic since many bindings
can be applied to an interface >and there is not necessarily even a
<service> element to list the bindings that are provided. I would >therefore
maintain that the @binding is required.

I just thought that it might be possible to keep an association between
bindings and interfaces, even if a service element is not available.
In that case, if @binding is absent, then if there are multiple bindings
available, a binding of the same type as that of the binding  for the
referencing service must be chosen; otherwise, a single available binding is
used.
If a specific binding is required, then @binding may/should be specified.
Alternatively, some client runtimes may use some policy/configuration info
when choosing between multiple bindings.

Cheers
Sergey Beryozkin
Zandar Technologies, Dublin, Ireland

----- Original Message -----
From: Arthur Ryman
To: Sergey Beryozkin
Cc: Amelia A. Lewis ; WS-Desc ((Public)) ; www-ws-desc-request@w3.org
Sent: Monday, April 28, 2003 3:00 PM
Subject: Re: proposal for restricting a service to a single interface



Sergey,

Yes, the default value for @xpath is "." which selects the entire part.
That's in the proposal.

We could also make @part optional if there was exactly one part in the
message.

Making the binding optional is much more problematic since many bindings can
be applied to an interface and there is not necessarily even a <service>
element to list the bindings that are provided. I would therefore maintain
that the @binding is required.

Arthur Ryman,

Received on Monday, 28 April 2003 11:58:43 UTC