On 9/19/07, Keith Kyzivat <[EMAIL PROTECTED]> wrote: > > Ahh sent before I completed my second thought.. > > On 9/19/07, Keith Kyzivat <[EMAIL PROTECTED]> wrote: > > > > > > > > btw, why the BridgeMixSilence.diff patch which to solve "not seeing > > > outbound rtp packet" mail has not been check in? > > > > > > I believe that was a quick hack that fixes it, but not in a very elegant > > way. > > In order to > > > > In order to be checked in, someone needs to re-write it to be a bit less > hackish. Alex and I don't have the time right now, but if someone else > wishes to take a stab at it, go right ahead. >
caveat: I'm not tremendously familiar with the bridge code Ahh I think I see the reason it is a hack and should be re-written. The problem that the patch corrects is due to what Alex thinks is some sort of 'internal DTX' in sipXmedaLib, causing audio buffers to indicate they are not active, when they really are. The solution was to eliminate a check that eliminated work when audio buffers were indicating they aren't active when mixing -- inactive buffers don't need to be mixed in. So ultimately this means that *all* buffers coming into the mixer (when in LINEAR_COMPLEXITY mode) get mixed together, whether or not they're set as active or not. Now, I'm not sure if the 'inactive' buffers just hold silence, or if they hold data that shouldn't be mixed. The former is not very desirable, but is fairly benign. The latter is not what is desired. So, that's my summary of the behavior of the problem and solution FWIW. -- Keith Kyzivat SIPez LLC. SIP VoIP, IM and Presence Consulting http://www.SIPez.com tel: +1 (617) 273-4000
_______________________________________________ sipxtapi-dev mailing list [email protected] List Archive: http://list.sipfoundry.org/archive/sipxtapi-dev/
