Alain Totouom wrote: > Hi Benny, > > Benny Prijono wrote: >> Alain Totouom wrote: >>> This assertion is only true during the call setup (*first* 18x|200-INVITE). >>> Afterwards you can't rely on that information for the duration of the call! >>> When a re-invite (hold e.g.) occurs the role might be different etc.. >> I can understand why you think call->role behaves like that, but >> actually that's not correct. In pjsip, call->role (or dlg->role and >> inv->role to be precise) should stay the same for the duration of >> the call. >> > > please correct me if I'm wrong: > isn't it just one of the "very" convenient and useful features of *pjsip*. > In my understanding of rfc.3261 ?17 the UAC and UAS roles are transaction-based > rather than dialog-based. That is the reason why I've stated before that > inbound/outbound was set by the very first INVITE-Transaction. True. In 3261, UAS and UAC role applies to transaction rather than dialog. For dialog, maybe I should have named it callee or caller instead, but since we already have pjsip_role_e I thought I just reuse it, hence the confusion! > But I'll give you the point if you say we're talking about the gender of > electrons here ;) Yeah. For dialog, invite session, and call, UAC means caller and UAS means callee. And this role stays the same throughout the duration of the call. -benny >> regards, >> -benny
