[Ekiga-devel-list] More analysis of Alsa vs Pulse

[Andrea <mariofutire googlemail com>]

Hi,

After changing the uninitialised values I've kept comparing the behaviour of ekiga with ALSA and
Pulseaudio.

I've thought of reporting here what I have found in the hope that someone could have a better
understanding and maybe identify the core of the issue to report to PulseAudio.

I am analysing what happens when I play the ringtone via the conf dialog.

Basically in PSoundChannleALSA there is a method "Open()".
It is called twice, the first time it gives slightly different results (even after correcting the
uninitialised), but the second time it gives the same values

2009/01/26 20:39:14.818	  0:10.651	AudioEvent...0xb6804b90	Alsa	Buffer time is 20000
2009/01/26 20:39:14.818	  0:10.651	AudioEvent...0xb6804b90	Alsa	Period time is 5000

Afterwards the actual audio output starts. This is done via Write().

What happens when using the pulse-alsa module is that there are many underruns after calling
snd_pcm_writei().

In order to fix the underrun snd_pcm_prepare is called and I think this is responsible for the very
bad quality AND for the fact that the stream volume controller in the PulseAudio Volume Manager
keeps disappearing and reappearing.
I've tried snd_pcm_recover() but there is no difference.

It seems that each time an underrun occurs, a nasty glitch can be heard.

I have no idea what causes the underrun.

Does anybody have a better understanding of all that?