[asterisk-users] Problems With Playback of Audio On SIP Only System

Kevin P. Fleming kpfleming at digium.com
Wed Jan 21 10:54:15 CST 2009


Mark Michelson wrote:

> I started thinking about this and I'm not sure how you can check for this at 
> configure-time or build-time. While it would be easy to check what gcc version 
> is being used, it is not so easy (dare I say, not possible) to see if 
> gsm-formatted sounds are going to be used in this setup. Besides the fact that 
> core sounds are not placed on the system until make install is run, we can't 
> know if there are out-of-tree gsm-formatted sound files which will be used, too. 
> Did you have something clever in mind for such a check?

I would approach this an entirely different way: if the system exhibits
this problem, then building codec_gsm will produce a module that cannot
properly encode (or decode, whichever direction is the issue) GSM
streams. It does not matter whether the user plans to use GSM or not; if
we tell them their system cannot build a non-broken codec_gsm, and they
don't plan to use GSM, then they can disable it and not have to worry
about it. If, however, they plan to, or might, use GSM (whether it is
sound files or audio streams does not matter), then they will need to
address this issue or they could have audio corruption.

If we know which specific piece of assembly code is incorrectly compiled
or optimized in the GSM library and causes this problem, it is
conceivable we could write a configure script test that compiles that
same code, runs a test, and checks the output. We avoid writing
configure script tests that check for version numbers whenever possible,
it's much more reliable to check for the specific feature (or bug) that
we are concerned about, because distros frequently backport and
forwardport patches, so the version numbers become unreliable.

With all that said; if the current releases of Asterisk have a
functional workaround for this problem, and users who have this problem
can just update to the latest release and the problem will go away, then
we need to just write a specific announcement to that effect. The only
reason any configure script checking would be needed is if the problem
still exists and we cannot fix it in the Asterisk source code.

-- 
Kevin P. Fleming
Digium, Inc. | Director of Software Technologies
445 Jan Davis Drive NW - Huntsville, AL 35806 - USA
skype: kpfleming | jabber: kpfleming at digium.com
Check us out at www.digium.com & www.asterisk.org



More information about the asterisk-users mailing list