[asterisk-dev] sip/rtp jitterbuffer in 1.4? (Chicken or the egg?)

Andrew Kohlsmith akohlsmith-asterisk at benshaw.com
Tue May 30 08:59:14 MST 2006


On Tuesday 30 May 2006 11:43, Jared Smith wrote:
> If you want a solid reliable RTP jitterbuffer in 1.4, then help out!
> Jitterbuffers don't invent themselves, and they don't stress-test
> themselves, and they don't debug themselves.  In short, it makes me sick
> to hear some developers blame other developers for not having done the
> work necessary for a solid reliable jitterbuffer.  If *we* really want a
> good jitterbuffer, then *we* (myself included) better stop sitting
> around wishing it would happen and actually *do the work* to get
> something in shape before 1.4 is upon us.  As far as I can tell, very
> few people have even tried the existing proposed jitterbuffer for 1.4.

I am guilty of this, but as Steve said, the code Just Did Not Work.  Fuck man, 
it didn't even build!  Not simple fixes to make it build either.

I am also guilty of confusing the rtp jitterbuffer branch with the test-me-now 
branch...  test-me-now had everything and the kitchen sink thrown in it -- I 
was simply not willing to test EVERYTHING at once on a production system, so 
I avoided it.

> I should also respond to the cry for putting the proposed jitterbuffer
> in trunk.  While I'm in no position to speak for anyone but myself, I'm
> quite sure that were the proposed jitterbuffer to go into trunk, we'd
> get a huge backlash from people running trunk complaining that suddenly
> their RTP audio is all messed up!  Seems to me like we can't have it
> both ways...

Nope.  I run trunk.  Whenever something changes and it breaks my setup I bitch 
about it a little in -dev but everyone there who sees me in there practically 
every day knows damn well that I take responsibility for running trunk.  If 
something breaks in an update, I roll back and reinstall to avoid the 
breakage, and I work on the broken stuff in the non-production environment.  
That way I get the best of both worlds.

Anyone running trunk and seriously bitching that it's not staying still has 
got some serious issues with their cognitive processes.

-A.



More information about the asterisk-dev mailing list