kiax 0.8.51 - please review & upload

[George Danchev]

On Monday 26 June 2006 00:58, George Danchev wrote:
> On Sunday 25 June 2006 18:28, George Danchev wrote:
> --cut--
>
> > > I can read on my tests a filter flag of 15 prior to the segfault,
> > > now... I'd like to know why it is segfaulting, I mean, I would expect
> > > it not to work well, but not to segfault, I don't know why this
> > > happens, could this be that kiax is only implemented to use the echo
> > > cancellation of it's modified nonfree stuff?
> >
> > Right, that is the case here. I think the problem is at our libspeex
> > packages, e.g. if you diff mdf.c implementations kiax has and our
> > libspeex has, you will find at least one different expectation in
> > speex_echo_cancel function (last argument):
>
> ok, since I didn't managed to track the segfault issue down ( which seems
> to be somewhere in our existing speex or iaxclient packages), I added a
> patch to kiax packaging (upon other existing patches ;-) to revert dfsg
> repackaging to kiax/lib/ with lib/iLBC, lib/CVS, lib/aec_nlms removed. The
> rest of the stock kiax/lib if LGPL code. If you want to revert that patch,
> please do:
>
> from kiax/trunk
> patch -R --dry-run -p0 < ../ec-enable.patch
>
> If that approach proves to be the right thing [tm] to go, we will apply
> that to trunk as well. At least segfault is gone when EC is being enabled
> in kiax.

I forgot to mention that kiax (0.8.51.dfsg-2) is currently ready for upload, 
if no objections to the above mentioned repackaging/linking change arise of 
course.

btw, be aware of 'tar 1.15.91 breaks lintian and linda' issue - #375300. So 
grab an old tar package from snapshot.debian.net... tar=1.15.1-6 works for 
me.

-- 
pub 4096R/0E4BD0AB 2003-03-18 <people.fccf.net/danchev/key pgp.mit.edu>
fingerprint 1AE7 7C66 0A26 5BFF DF22 5D55 1C57 0C89 0E4B D0AB