Re: [wsrm] Question on wsdl operation support

[Tom Rutt <tom@coastin.com>]

Sunil Kunisetty wrote:

>  
>
> Tom Rutt wrote:
>
>> the current spec states the following: in lines 751 thru 757:
>> "
>> 2.9 WSDL Operation Type
>> This specification supports Reliable Messaging capabilities for WSDL 1.1
>> [WSDL 1.1] One-way
>> and Request-response operation types only. While a Request-reponse
>> operation can use any of
>> the three RM-Reply patterns to receive acknowledgments or faults, an
>> One-way operation can
>> only use either Callback or Poll RM-Reply pattern. See the table below
>> for a complete support
>> matrix:
>> "
>>
>> This, and the table, are not correct.
>>
>  A _WSDL 1.1 R-R operation _can still use any of the 3 patterns it's 
> just that
>  it doesn't make much sense to use the Callback and Poll pattern. 
> Remember,
>  we initially had the value *Allowed* under Callback and Poll pattern 
> for WSDL
>  1.1 R-R operation. Later we changed all 3 of them to be *Supported* 
> and wanted
>  to have a footnote saying that although an user can use either the 
> Callback or
>  Poll pattern for a WSDL 1.1 R-R operation, it is encouraged to use the
>  Response Reply pattern itself so to minimize the network round trip.

OK,

>>  
>>
>> A wsdl RequestResponse operation could have a payload in the response.
>> If there is a payload,  the poll response has no way to return a batch
>> of payloads.
>>  
>>
>  If the user selects a Poll pattern with R-R, the response to the 
> actual operation
>  will come in the Transport (Http) response and this response doens't 
> contain
>  any RM Headers. When a subsequent Poll request is sent, then a Poll 
> response
>  is sent back. So the application payload doesn't overlap with the 
> Poll response.

We should clarify this.  I now agree that one could use the callback or 
the poll response for the
reliability ack with wsdl request/response.

>>  
>> Also, if we want to batch acks , then we have the same problem, in that
>> we cannot return batched response payload from the wsdl
>> requeste response.
>>  
>
>  Batching doesn't make sense for Response pattern case as the 
> definition says
>  that the ack MUST be sent back with the underlying transport and also 
> notion
>  of the Sender is not clear in this case.

I think the RM-fault for response reply pattern currnelt map onto soap 
fault, this means that
the batching of faults cannot be done.

I agree the protocol will be simpler if our response reply pattern does 
not allow batching.,

>  Batching for Callback pattern is based on the replyTo endpoint.
>  
>  Batching for Poll is based on the request (i.e. if the poll request 
> itself has
>  a SequenceNumberRange).
>
>>  
>> Thus I thinke we have to change the table to have callback and polling
>> not supported for Request/response wsdl operations.
>>  
>
>  Change is not necessary as explained above.

I now agree, perhaps a clarfication would be useful.

>>  
>> Tom Rutt
>>
>> -- 
>> ----------------------------------------------------
>> Tom Rutt                email: tom@coastin.com; trutt@fsw.fujitsu.com
>> Tel: +1 732 801 5744          Fax: +1 732 774 5133
>>
>> To unsubscribe from this mailing list (and be removed from the roster 
>> of the OASIS TC), go to 
>> http://www.oasis-open.org/apps/org/workgroup/wsrm/members/leave_workgroup.php.
>>


-- 
----------------------------------------------------
Tom Rutt		email: tom@coastin.com; trutt@fsw.fujitsu.com
Tel: +1 732 801 5744          Fax: +1 732 774 5133