In line.

"Francois Audet" <[EMAIL PROTECTED]> wrote on 07/18/2007 11:26:54 AM:

> > > What I mean is that the error response may not be delivered 
> > to the UAC 
> > > because a forking proxy will only deliver one final 
> > response. If there 
> > > are multiple error responses on the individual forks, one being the 
> > > one with the RPH, and the other being some other response, 
> > it is up to 
> > > the proxy to decide which one to forward back.
> > > 
> > > So the UAC can not rely on receiving the RPH to "re-intiate" the 
> > > session or provide some alternative treatment. This may or 
> > may not be 
> > > a problem depending on the application (that's why I said "if it 
> > > absolutely has").
> > > 
> > 
> > I don't think that will be a problem. 
> > 
> > For one thing (for the namespaces I am working with) the 
> > expectation would be that ALL the responses (except 100 and 
> > 403) would have the RPH on the response, so it wouldn't 
> > matter which one made it back to the UAC. 
> 
> I am not talking about multiple provisional responses from the same
> endpoint.
> 
> I am talking about multiple responses (provisional and final) from 
> multiple forks. For example, multiple 4XX responses.

Yes, I understood that.
> 
> While it is possible that a single forking proxy may decide to 
> include the RPH in all responses, that proxy has absolutely no 
> control over downstream and upstream proxies who may not include it
> (they may be proxies in different domains).
> 
> That's the point I'm trying to make. You can not enforce the 
> presence RPH in responses in a global system.

Agree. For the context I am working in, a "global system" is not of 
concern.  I would not expect universal "RPH in responses" in a global 
system.

> 
> > Secondly, the expectation is that the devices responsible for 
> > "re-initiating" WOULD be stateful, and not dependant on the 
> > information in the RPH in the response.  It is the 
> > intervening devices that might not be stateful, and would 
> > take advantage of the RPH information in the responses. 
> 
> That is precisely the type of clarification that needs to be included
> in the draft. It points out the constraints of the mechanism so 
> implementors use the feature properly.
> 
> Some text explaining that the UAC needs to be able to deal with the RPH
> not being present in responses gracefully (and why) is needed.

I am not sure why that needs to be in the ID.  It is an implementation 
specific issue. 

Since  RFC 4412, as currently published, forbids RPH in responses, I think 
it "goes without saying" that "the UAC needs to be able to deal with the 
RPH not being present in responses".

Further, while I know that the systems (and namespaces) _I_ work with 
intend to use RPH in responses, I also have been told that other systems 
(using other namespaces) do not intend to put RPH in ANY responses.  I do 
not think there is any need to standardize this aspect, just to adjust the 
standard so it is PERMITTED.
> 
> Similar text for proxies also need to be there. What if a proxy 
> forwards a request with RPH and it receives nultiple responses from
> the downstream proxies, with some including the RPH and others not
> including it? Does this need to be standardized?

Again, I do not think there is a need to standardize this aspect.

Janet
> 
> 
> _______________________________________________
> Sip mailing list  https://www1.ietf.org/mailman/listinfo/sip
> This list is for NEW development of the core SIP Protocol
> Use [EMAIL PROTECTED] for questions on current sip
> Use [EMAIL PROTECTED] for new developments on the application of sip
_______________________________________________
Sip mailing list  https://www1.ietf.org/mailman/listinfo/sip
This list is for NEW development of the core SIP Protocol
Use [EMAIL PROTECTED] for questions on current sip
Use [EMAIL PROTECTED] for new developments on the application of sip

Reply via email to