Re: Issue i020, subissue 3 proposal

Tom Rutt wrote:
> 
> Anish Karmarkar wrote:
> 
>>
>> Per my AI, Paco and I discussed subissue 3 of issue i020 last night 
>> and a proposed resolution for it. Here is what we would like to 
>> propose to resolve subissue 3.
>>
>> [[ Please note that Paco and I have only generally agreed on what the 
>> resolution should be; he has not seen the wordings in this email and 
>> therefore it should not be assumed that he has agreed to the wordings ]]
>>
>>
>> Subissue iii [1] is:
>>
>> -----
>> An EPR allows one to include (optionally) a service endpoint/port.
>> If such an endpoint/port is included in an EPR, what is the relationship
>> between the value of the [address] property and the URI value in the
>> [service-port] property? We have said that the [address] property is a
>> logical address and not necessarily the physical endpoint where messages
>> can be sent and how the mapping between logical to physical takes place
>> is an extensibility point. Is that true if a service QName is present in
>> the EPR. I.e., should our spec say that if the service QName is present
>> then the physical address is what is specified by the wsdl port.
>> -----
>>
>> Proposed resolution:
>>
>> 1) When the EPR minter includes a [selected interface], and/or 
>> [service endpoint] then the EPR is considered to be specific to the 
>> [selected interface] and/or [service endpoint]
>>
>> 2) When an EPR contains [service endpoint] with a QName identifying 
>> the service element and an NCName identifying port/endpoint, then the 
>> information specified in the port/endpoint (including the network 
>> endpoint address) is used to send messages to the endpoint 
>> identified^h^h^h^h^h^h^h^h^h^h specified by the EPR. I.e., the 
>> physical address/binding used to contact the endpoint is the one 
>> specified in the port/endpoint. This physical address may be the same 
>> as the one in the [address] property. If it is different from the 
>> value in the [address] property then the [address] property is 
>> considered to be a logical address.
> 
> 
> I really like proposal 2) above.  However, based on discussion at the 
> Teleconference I would like
> to add a clarification regarding the use of Physical address in the text:
> 
> Each specific protocol has its own "definition" of physical address.
> 
> For the Soap/HTTP Post binding, this text is meant to refer to the the 
> HTTP "address", which is
> the request-uri value in the POST request-line (rfc2616 section 5.1) 
> along with the value for the HOST Line in the request-header (rfc2616 
> sections 5.3 and 14.23)..
> .

Could we just say that the binding defines how the address is dereferenced?

> If the word "physical address" is inappropriate to refer to this "http 
> address" concept then we
> should come up with a better term.
> 

My proposal was not meant to be included as is. It was meant to convey 
the intention.

I'm not sure there is a benefit to creating distinctions between 
physical/logical address. I have filed an issue regd this.

> Tom Rutt
> Fujitsu
> 
>>
>> 3) When an EPR contains [service interface] property with a QName 
>> identifying the service element but an NCName identifying 
>> port/endpoint is *not* specified, then the information specified in 
>> any of the port/endpoint (including the network endpoint address) that 
>> implements the [selected interface], if present, is used to send 
>> messages to the endpoint specified by the EPR. I.e., the physical 
>> address/binding used to contact the endpoint is the one specified in 
>> any of the port/endpoint. The [address] property is considered to be a 
>> logical address if there are more than one ports/endpoints defined in 
>> the service element.
>>
>> Comments?
>>
>> -Anish
>> -- 
>>
>> [1] 
>> http://lists.w3.org/Archives/Public/public-ws-addressing/2005Jan/0101.html 
>>
>>
> 
> 

Received on Monday, 21 February 2005 19:37:08 UTC