[Asterisk-Users] Crazy stuff in latest CVS HEAD

[steve at daviesfam.org]

On Tue, 19 Jul 2005, Andrew Kohlsmith wrote:

> On Tuesday 19 July 2005 08:28, David Burgess wrote:
> > I am new to this list. What is the process of reporting something that is
> > completely busted? Or do we just patiently wait?
> 
> If you're running CVS HEAD be advised that this is the live development code, 
> it mostly works most of the time but little inconsistencies can and do 
> inevitably creep in as the code evolves.


Hi listees,

I have been having a problem that my box is generating insane outgoing 
iax2 timestamps.  These completely freak the jitterbuffer etc.

I reverted my channels/chan_iax2.c back to version 1.319 (cvs update -r
1.319 chan_iax2.c) and the timestamps on an IAX2 call seem after a quick
test to have come right or mostly right.

This reverts all the changes to chan_iax2.c in respect of time
calculations that went in with bug 4504.

So - it seems that this patch is responsible.

Now - is it just me, but this is a 92k patch that touches 38 Asterisk 
source files.  That's pretty invasive.  And it seems we are in the runup 
to a 1.2 "stable" release, a time when paranoia and extra care should be 
taken to avoid unnecessary code changes.

How come this goes into CVS-HEAD without a single note on the bug that it
has been applied by someone and works for them?

Given that people are having trouble in SIP and IAX2 and others, I presume
one of the new functions is broken, or alternatively that there could be
some bug in some gcc versions to do with passing structs by value?

(mine is gcc 3.3.5-20050130)

This patch went in on 05-07-15 so perhaps a checkout from before that will 
help other people  (cvs checkout -D 2005-07-14 asterisk ...)

Regards,
Steve