< Previous by Date Date Index Next by Date >
< Previous in Thread Thread Index  

Re: [reSIProcate] Please help - Register is Aborting.


After checking some traces the tid's were the same. However, the latest
code does not have this problem.

Thanks for the information.

Brian.

On Thu, 2004-07-08 at 19:06, Alan Hawrylyshen wrote:
> On Jul 8, 2004, at 15:36, Brian D'Souza wrote:
> 
> > Since, the transaction id
> > for #34 and #35 are the same, the TransactionState class matches
> > REGISTER #35 to the transaction for Register #34.
> 
> Are you implying that message 35 is from the DUM and has the same tid?
> That's definitely a problem and will cause all sorts of grief.
> Messages send to the stack should not have the branch-parameter in the 
> Via set manually.
> The stack will set it for you. If you need to put something in the 
> branch-param, please use the clientData() methods.
> 
> I also recently mentioned to Jason that I discovered a race condition 
> (as yet unverified) in the code that could potentially cause problems 
> if the retransmission timer fires (or more specifically is serviced) 
> prior to the DNS resolution completing. The timer may need to be 
> started elsewhere.
> 
> Do either of these suggestions help you understand what might be 
> failing? Please let us know.
> If your application is sufficiently small, it might be worth having 
> someone who's been in the stack a whole lot take a look.
> 
> Alan Hawrylyshen
> Jasomi Networks Inc.
> http://jasomi.com/
> a l a n a t j a s o m i d o t c o m
> 
>