[Asterisk-Users] Downgrading Asterisk

[Rich Adamson]

> > It seems the choppy (and almost unusable) audio in Head is only impacting
> > "some" cisco users, and since these problems are not impacting
> > the few that
> > can read code, use cisco phones, and are impacted, we're stuck with the
> > problem. The problem seems to be very evasive, however switching the iax2
> > links to use only iLBC (and not gsm) has corrected issues for some.
> 
> In my case I never had any problems doing SIP to SIP - even from non-cisco
> to the cisco phones or visa versa.  I only had problems with calls between
> the cisco and the POTS world through chan_capi.  This problem has been
> solved by the patch to rtp.c posted in this mailing list about a month ago
> (an if statement commented out).  With that patch HEAD works perfectly ok
> for me (ahem - I _hate_ writing that).

Lars,

By commenting out that if statement, you are disabling the function that ties
the timestamps together, and will have to keep doing that every time you
update code. That's a problem "bypass" and not a real "fix". (Also, if
you're not very carefull with that, cvs update will likely fail to 
update the rtp.c code.)

I've asked several times why it's important to tie the timestamps together,
and as of today no one has hinted or even guessed at the reason for
it. Therefore, pure guess is that Mark has some architectual reason 
(probably related to datastream distortion) for it that will bite us
later on.

Since I'm not a programmer and can't read the code worth a damn, I'd have
to venture a guess that iax2->rtp has had many issues addressed, but other
channels such as capi->rtp still need to be cleaned up. Since the noise
level seems to be oriented around iax2 and capi, its probably fair to 
assume other channels are working as expected.

Rich