[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Warzone 2100 packaging status report

Hash: SHA1

Eddy Petrișor wrote:

>> otherwise runs fine (as long as nothing gets damaged).
>> I did have time to do the bughunting yet, but I prepared the updated
>> packages (in SVN). Maybe we should upload them to experimental? The game
>> is perfectly stable on 32-bit systems after all.
> That seems sane enough... a later upload might make sense in sid and we could submit an RC bug ourselves to prevent
> migration to testing until we make it work. I consider this a better approach than removing non-working arches since we
> would have a way to say "please try now" instead of "I think we fixed it, but we must do another upload with your arch
> enabled" (i.e. "some testing" as opposed to "we do the testing ourselves")

After hours of black magic I am almost sure this is caused by OpenAL. We
do not have dbg libs for that, so I cannot yet say for sure, but it does
work just fine without sound.

The backtrace of the crash seemed totally random at first - as if some
random memory address was nuked. But after some time I noticed that this
has higher possibility of occurin in either OpenAL, or fglrx. And by
eliminating fglrx, only OpenAL is left. I also noticed that
self-compiled version of TORCS 1.3.0 is also very crashy with OpenAL
sound driver on my system.

Here are the results of my experiments:

      |  sound  | no sound
- ------+---------+----------
fglrx | crashes |   runs
mesa  | crashes |   runs

Before I realized that this may be a bug of our own, I posted this [1]
to Warzone developer mailing list, but nobody replied yet.

I would really appreciate some testing here. If this only happens on a
small number of systems, there is no reason to hold this back anymore.
The packages are ready - it is only a matter of grabbing the latest SVN,
and putting it together.

[1] https://mail.gna.org/public/warzone-dev/2007-06/msg00049.html

Version: GnuPG v1.4.6 (GNU/Linux)


Reply to: