Re: "interface" attribute info item on service component

Ram,

Point (3) is:

Moreover, if the service component has an interface attribute info 
item that extends from two other interfaces, can the endpoint defined
within it refer to bindings that were defined for the parent
interfaces ? If yes/no, should this be reflected in the core language
spec ?


The spec requires that the binding of an endpoint must either be 
interface-less, or ref to the same interface as the enclosing service. As 
I recall, the alternative got too complex.

The following diagram might help:



Arthur Ryman,
IBM Software Group, Rational Division

blog: http://ryman.eclipsedevelopersjournal.com/
phone: +1-905-413-3077, TL 969-3077
assistant: +1-905-413-2411, TL 969-2411
fax: +1-905-413-4920, TL 969-4920
mobile: +1-416-939-5063, text: 4169395063@fido.ca



"Ramkumar Menon" <ramkumar.menon@gmail.com> 
Sent by: www-ws-desc-request@w3.org
05/29/2006 11:41 PM

To
Arthur Ryman/Toronto/IBM@IBMCA
cc
www-ws-desc@w3.org, www-ws-desc-request@w3.org
Subject
Re: "interface" attribute info item on service component






Hi Arthur,
 
Thanks for the detailed explanation.
Maybe I was not clear in my question (1) .
I was explicitly referring to the "interface" attribute on the <service> 
nodes. Woudnt the interfaces that are referred to as attributes on the 
<service> nodes need to have atleast one operation within them, either 
declared / inherited ? Would it make sense to explicitly state this in the 
spec ? 
Similarly, for the second question, I was referring to those bindings that 
are referred to from within the <endpoint> node as an attribute - wdnt 
these referred bindings need to be referring to an interface mandatorily ? 
Again, if it makes sense, would it better if we explicitly state ithis in 
the spec ? 
 
I would also appreciate your thoughts on point (3).
 
Thanks again!
 
rgds,
Ram

 
On 5/29/06, Arthur Ryman <ryman@ca.ibm.com> wrote: 

Ram, 

It might be useful to have an interface that just defined faults, so -1 to 
requiring one or more operations. 

An endpoint refers to a single binding. If the binding refers to an 
interface, it must be the same as the service's interface. Note that 
generic "interfaceless" bindings are possible. 

Arthur Ryman,
IBM Software Group, Rational Division

blog: http://ryman.eclipsedevelopersjournal.com/ 
phone: +1-905-413-3077, TL 969-3077
assistant: +1-905-413-2411, TL 969-2411
fax: +1-905-413-4920, TL 969-4920
mobile: +1-416-939-5063, text: 4169395063@fido.ca 


"Ramkumar Menon" <ramkumar.menon@gmail.com > 
Sent by: www-ws-desc-request@w3.org 
05/23/2006 02:36 PM 


To
www-ws-desc@w3.org 
cc

Subject
"interface" attribute info item on service component










Three fundamental questions.

Would it be useful to add a clause for the <service> component stating
The "interface" attribute information item should point to an 
interface that has non zero number of "operation" element information
items within it.
If not, we cd as well have service components that could possible be
empty, and allow them to extend other service components, reflecting 
the same semantics we have defined for interface inheritance -
considering that one service component is related to exactly one
interface.

Am I right if I state that if all "binding" attribute info items that 
had been defined on the endpoint node should have been associated with
an  "interface" attribute information item? What does it mean to be
otherwise ?

Moreover, if the service component has an interface attribute info 
item that extends from two other interfaces, can the endpoint defined
within it refer to bindings that were defined for the parent
interfaces ? If yes/no, should this be reflected in the core language
spec ?

rgds,
Ram
-- 
Shift to the left, shift to the right!
Pop up, push down, byte, byte, byte!

-Ramkumar Menon
A typical Macroprocessor





-- 
Shift to the left, shift to the right!
Pop up, push down, byte, byte, byte!

-Ramkumar Menon
A typical Macroprocessor