Tricky situation. I think UAS should accept ACK and send BYE immediately. In UDP we can ignore but in TCP we can't ignore.
-----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Brett Tate Sent: Wednesday, June 18, 2008 6:09 PM To: Paul Kyzivat; [email protected] Subject: Re: [Sip-implementors] Call does not go through if ACK is receivedwithContent-Length more than the actual body size > While you could simply ingore the ACK, that probably isn't helpful in > getting things working. I don't disagree; however my understanding is that it has the same benefit/curse of discarding responses. Did rfc3261's recommendation concerning discarding responses have anything to do with UDP delivery issues potentially resolvable by a retry? > I'd be inclined to just pad out the message to the specified length. > That then may result in a body that is incorrectly formatted. > But for ACK that will only be the case if there is an ansswer in the > ACK. If so, you can then cope with the errors found in the body. > > But generally this is an error case that you can cope with as find > works best for you. > > Paul > > Brett Tate wrote: > >> Please can you tell me the behavior of UAS in the following case: > >> If UAS is getting the ACK with content-length more than the actual > >> body length and all the headers are correct? > > > > See rfc3261 section 18.3. However since there is no > response for ACK, > > the "SHOULD generate a 400" does not apply. The ACK should be > > discarded similar to what occurs for responses. > > > > _______________________________________________ > > 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 _______________________________________________ Sip-implementors mailing list [email protected] https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors
