rhythmbox segfault / sound quality problems in fc4t3...

[Chris McDonough]

On Wed, 2005-06-08 at 10:15 -0400, Bill Nottingham wrote:
> Chris McDonough (chrism at plope.com) said: 
> > Ah ha! I fired up gconf-editor and lo and behold
> > system/gstreamer/0.8/default/audiosink was set to "esdsink".  I changed
> > it to "osssink" and it sounds beautiful now.. great.  Thanks for the
> > direction!  I suppose I should look into the benefits of running esound
> > and esdsink instead.
> 
> esd has known issues with the dmix config used in FC4t3; since dmix
> does the mixing for you, there's really no reason to use esd as a
> gstreamer output.

Thank you, that's useful to know.

> What's best is to set it to alsasink with gstreamer-properties.

Ah.  Duh.  I didn't know about gstreamer-properties (it doesn't seem to
have made its way into my GNOME menus and I'm getting so
pointy-and-clicky lately... ;-)

Setting it to osssink works fine and the pipeline test passes.  If I set
it to alsasink, and I run a pipeline test from within
gstreamer-properties, it fails with:

Failed to construct test pipeline for 'ALSA - Advanced Linux Sound
Architecture'

This appears to be the same symptom noted in
https://www.redhat.com/archives/fedora-test-list/2004-October/msg01258.html .

Colin Walters apparently merged some related fixes into rawhide in
mid-2004:

https://www.redhat.com/archives/fedora-test-list/2004-July/msg00270.html

But here's what happens when I try to use alsasink from gst-launch:

gst-launch filesrc location=01\ -\ South\ Of\ Heaven.ogg ! spider !
alsasink
RUNNING pipeline ...
ERROR: from element /pipeline0/spider0/vorbisdec0: Internal GStreamer
error: pad problem.  File a bug.
Additional debug info:
gstpad.c(2562):
gst_pad_set_explicit_caps: /pipeline0/spider0/vorbisdec0:
failed to negotiate (try_set_caps with "audio/x-raw-float,
rate=(int)44100, channels=(int)2, endianness=(int)1234, width=(int)32,
buffer-frames=(int)0" returned REFUSED)
Execution ended after 16 iterations (sum 358213000 ns, average 22388312
ns, min 24000 ns, max 161889000 ns).

Obviously RB fails with the same sort of error.  I suppose I should just
file a bug with this info for the gstreamer folks (unless one already
exists), as it says?

- C