On Fri, 2009-04-03 at 13:39 -0500, Adam Roach wrote: > I agree that a UAC can't *count* on a non-200 response stopping the > provisional retransmission, since the error may have come from an > intermediary. Of course, if the error has an obvious recovery path (e.g. > 407), then the UAC should attempt recovery (largely because it doesn't > know whether the error came from the UAS or from an intermediary). > > However, if the PRACK makes it all the way to the UAS, reliability has > been achieved. Continued retransmission of the provisional response by > the UAS at that point provides no benefit. It does, however, consume > gratuitous bandwidth. (The UAS does need to deal with the possibility > that the UAC will re-attempt the PRACK, however). Adam's response seems to be the obvious solution to this problem. The only thing I would clarify is that if the UAC sends a PRACK which receives a 4xx response, and the UAS *didn't* send the response, the UAC will discover this directly by receiving another copy of the reliable provisional response, and then the UAC will then know that it has to send a new PRACK acknowledging that reliable provisional resopnse. I admit that this may have complicated interactions with the offer/answer model, but otherwise, it seems straightforward, and indeed, it seems to be the only solution. But perhaps I am overlooking something. Dale _______________________________________________ Sipping mailing list https://www.ietf.org/mailman/listinfo/sipping This list is for NEW development of the application of SIP Use sip-implementors@xxxxxxxxxxxxxxx for questions on current sip Use sip@xxxxxxxx for new developments of core SIP
