On 4/12/2011 12:21 PM, Randell Jesup wrote: >> 3 apr 2011 kl. 13.23 skrev Iñaki Baz Castillo: >> >>> 2011/3/31 Olle E. Johansson<o...@edvina.net>: >>>> If you are sending only ringback, I would recommend sending 180 with SDP >>>> instead of 183. If you're sending 183, I can't move my state machine to >>>> ringing state, which would help a lot of 3rd party apps. If you send >>>> ringback in 183 - they won't notice the ringing and would see the call >>>> going from calling state directly to answer, which is confusing. Sending >>>> ringback with 183 and not sending 180 is a problem for many apps. >>>> >>>> Use 183 only if you have an operator message to play. Otherwise, add SDP >>>> to the 180 Ringing. >>> >>> Hi Olle. So you propose that ringback (just pure ringback with no >>> voice announcement) should be always sent in a 180 response (with or >>> without SDP, as usually SIP phones generate 180 with no SDP, of >>> course). >> Yes. I am not saying that you should always send ringback with 180, but >> if you do want to send ringback in audio, use 180. > > I should tell you that this is opposite how almost all commercial > products work; 180 is normally used without SDP for telling the caller > to generate internal ringback, and 183 with SDP is used for providing > remote ringback (usually from a SIP->PSTN gateway). Most UA devices > don't send ringback themselves and just send 180 to the server. > >>> And just use 183 (always with SDP for sure) when the media contains an >>> announcement ("the number you are calling is not available" and so). >> exactly. >> >>> Am I right? >> Yes. >> >> 183 doesn't say much about the state change, but 180 is actually a >> state change indicating that something is alerting the target about an >> incoming call. This is very important in gateway situations, like a >> b2bua like Asterisk. > > A quote from this list from 2008 (use the archives!!!): > > RFC3960 gateway model gives guidelines about the, let's say, "correct" > behavior. The main rule says more or less "media has always the > precedence". > About the scenario you mentioned in the text, i.e. 183/SDP and then 180 > (no > SDP I assume) you should play a local ringback ONLY in the case you don't > receive any media (despite the media channel has been established). > In the reverse case, i.e. 180/noSDP and the 183/SDP the rule is the same, > meaning you should play the media. Should the media missing, you would > continue to play the local ringback due to the 180. > > This is what RFC3960 says. In the real world (especially when you > interoperate to the PSTN) it doesn't work in some case. My > recommendation is > to always use the latest received message as the most significant. Then, > in > the first case once receiving the 180 I would suggest to play a local > ringback stopping playing the media, in the second one to stop the local > ringback and play the media (if any). Using this approach, I didn't find > yet > any failing case when interoperating with PSTN (this is the most complex > scenario usually) in the operator's network I'm working with.
I don't understand. What is there about the "gateway" model that doesn't work in the real world with the PSTN? Thanks, Paul _______________________________________________ Sip-implementors mailing list Sip-implementors@lists.cs.columbia.edu https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors