[linux-audio-user] sndpeek: real-time 3D FFT + audio visualizatoin

[Ge Wang]

Sorry for the multiple post:

For those systems that are crashing somewhere in the
display driver (according to gdb), try the following:

     set LIBGL_ALWAYS_INDIRECT environment variable to 1

this disables hardware rendering.

if it works:
     - the problem is likely with the display driver
     - it will be very slow

if it doesn't:
     - there is a different bug in sndpeek we need to fix

Best,
Ge!


On Oct 27, 2004, at 6:19 PM, Ge Wang wrote:

> Thanks to all who have downloaded and sndpeek'd -
> and for the great suggestions, most of which we will
> put into the next version.
>
> regarding sndpeek crashing using JACK:
>
> We are trying to track this down.  So far ALSA seems fine -
> please let us know if sndpeek crashes with ALSA on your
> system.  There seems to be some issues with opening audio
> input using JACK with RtAudio on certain systems.  When
> this fails, there should be an exception thrown, but we found
> that this is disabled in the current build.  It is a possible source
> of the crash.  Stack traces suggest sndpeek crashes deep in
> the glutCreateWindow().  Perhaps updating the video driver
> may fix this problem.  We hope to have a more robust fix soon!
>
> Best,
> Ge!
>
> On Oct 26, 2004, at 10:51 PM, Kevin Ernste wrote:
>
>> Thanks Dan.  No joy after setting LD_ASSUME_KERNEL, same behavior.
>> Hopefully some of the bugs will get ironed out (here or with sndpeek),
>> it looks promising indeed.
>>
>> Kevin
>>
>> On Tue, 26 Oct 2004 23:00:56 +0100, Dan Mills
>> <dmills at spamblock.demon.co.uk> wrote:
>>> On Tuesday 26 October 2004 22:17, Fernando Pablo Lopez-Lezcano wrote:
>>>> On Tue, 2004-10-26 at 13:48, Kevin Ernste wrote:
>>>>> It runs here fine with ALSA (looks great, by the way, thanks for 
>>>>> the
>>>>> nice work!).  However, the JACK version locks up the terminal and 
>>>>> the
>>>>> process cannot be killed. in fact "ps" dies even trying to view the
>>>>> PID, etc.  I would be happy to try and better diagnose it if others
>>>>> have a similar problem.
>>>
>>> The jack version is OK here (recent CVS jack), radeon 9200 graphics, 
>>> fun
>>> program.
>>>
>>> Now if I just knew enough open GL to add calibration scales for the 
>>> frequency
>>> response graph life would be good!
>>>
>>> The ps dying thing is reminiscent of the problems qjackctl used to 
>>> have with
>>> the NPTL threading library on kernel 2.6, possibly try the 
>>> LD_ASSUME_KERNEL
>>> workaround?
>>>
>>> Regards, Dan.
>>>