Re: Choices for CR 20

Mark Baker wrote:

>As this relates to the TAG's endPointRefs-47 issue, I'd suggest that
>an absent wsa:To header should imply that the endpoint address is that
>provided in the envelope of the containing application protocol (when
>one is in use).  e.g. the HTTP Request-URI on an HTTP hop.
>
>Mark.
>  
>
I would see this as applying to CR 18, not CR 20.  CR 20 is a purely
syntactic question: If you don't see wsa:To, what value (if any) do you
assign to [destination]?  CR 18 is a semantic question of what do you
/do/ when you see anonymous in the [destination] (regardless of whether
wsa:To was anonymous or missing).

With that in mind, I believe this supports choice option 2 for CR 18
(allow anonymous in the [destination] for messages and define it as the
binding-specified destination of the message.)  I'm a bit unclear on why
HTTP in the usual case would be a "/containing application /protocol"
and not an "/underlying trans/* protocol" (your choice of "*"), but I'm
going to hope this is just a difference in terminology.

Note that option 2 is not completely consistent with our current
description of anonymous ("Some endpoints cannot be located with a
meaningful IRI; this URI is used to allow such endpoints to send and
receive messages. The precise meaning of this URI is defined by the
binding of Addressing to a specific protocol..").  It's consistent with
the second part (binding-specific), and it's consistent with the
approach we took for response endpoints (defining them in terms of
request-response), but it's generally not consistent with the first part
of the description (no IRI available).  Adopting option 2 would mean
deleting the (non-normative) first sentence in favor of giving a
somewhat precise meaning for the case of [destination].

>On 2/8/06, David Hull <dmh@tibco.com> wrote:
>  
>
>> There are basically three choices:
>>
>>
>>Status quo.  Missing wsa:To in the infoset means [destination] == anonymous
>>in the MAPs, always.
>>Limit this defaulting to the context of request-response.  If you want to
>>use an anonymous [destination] elsewhere, you have to do so explicitly.
>>Get rid of defaulting entirely.  You must always spell out what value you
>>want for [destination]. Separately from this, we can place various
>>restrictions on the use of anonymous [destination], however it may have
>>arisen, as part of resolving CR 18.  For example, in any of the three cases,
>>we could say that anonymous [destination] is only allowed for response
>>messages as a result of section 3.4.  We could also ban anonymous
>>[destination] altogether.
>>    
>>
>
>  
>