[sr-dev] git:master: modules/tm: new implementation of load_contacts()/next_contacts ()

[Alex Hermann]

On Friday 05 August 2011, Juha Heinanen wrote:
> Alex Hermann writes:
> > The issue is not with t_next_contacts, but with t_load_contacts. It is
> > not discarding the ruri of the request that made the request end up in
> > failure_route. Instead it is added again to the dset, where it obviously
> > doesn't belong.
> 
> after you have called t_load_contacts(), you should call
> t_next_contacts(), which does what i explained earlier.  after that ruri
> contains a contact with highest q value.
Of course, but is that happens to be the old ruri, it is used again directly. 

If it is not the highest q, it will be added to the contacts avp and it will 
be used again if t_next_contacts is called again.


> calling t_load_contacts()
> alone does not make sense.

I did not say that, i just said the issue is with t_load_contacts. 
t_next_contacts does function correctly, but as the old request uri is added 
to the contacts avp, some invocation (depending on ruri's q value) of 
t_next_contacts will show the failure of t_load_contacts.

-- 
Greetings,

Alex Hermann