" Think about this from the perspective of a UAS - if two UACs  
start talking to it at once and happen
to use the same branch parameter value, the results will be incorrect  
matching of message to transaction."

This is what we have experienced and hence I had asked this question.
Atleast from the RFC it was not clear to me that the "branch value MUST
be unique for all request sent by different UA". 
As Brett pointed out the, stack could have avoided this situation by
checking for sent by and the branch value for the transaction to message
matching but I do see that this is not followed by stack.  

Regards
-venkat 

-----Original Message-----
From: Robert Sparks [mailto:[email protected]] 
Sent: Friday, February 06, 2009 8:51 PM
To: Dale Worley
Cc: VYANKTESH TADKOD; [email protected]; Brett Tate
Subject: Re: [Sip-implementors] Question on Branch parameter usage

So I think we're identifying there is ambiguity in the sentence
"The branch parameter value MUST be unique across space and time for  
all requests sent by the UA. "
on page 39 of RFC 3261.

It was intended, and is necessary for correct operation, that this  
parameter be unique across space and
time. Think about this from the perspective of a UAS - if two UACs  
start talking to it at once and happen
to use the same branch parameter value, the results will be incorrect  
matching of message to transaction.

Unless someone objects, I'll create a bug to track this for  
clarification.

RjS


On Feb 5, 2009, at 5:07 PM, Dale Worley wrote:

> On Wed, 2009-02-04 at 14:04 +0100, VYANKTESH TADKOD wrote:
>> But in any case is it
>> acceptable to have 2 UAs generating the same branch value? As per  
>> RFC I
>> think the answer is "NO" but it is not very clear.
>
> As Brett noted, RFC 3261 seems to allow the branch to be only unique
> among all requests generated by a UA.  But the accepted practice is  
> that
> it must be unique over all space and time.  (Since the UA must be able
> to generate Call-IDs that are unique over all space and time, it can
> also generate branch parameters that are, so there is no additional
> burden on the UA.)
>
> Dale
>
>
> _______________________________________________
> Sip-implementors mailing list
> [email protected]
> https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors


_______________________________________________
Sip-implementors mailing list
[email protected]
https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors

Reply via email to