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

Esound versus Alsa versus GStreamer versus GNOME versus 2.6 versus ...



        Hi,

 How the hell is sound going to work in Etch under GNOME?

 1/ Current status

 At the lowest level:
 -------------------
 Sarge was bound to supporting OSS officially (in its kernels), and had
 optional ALSA packages for 2.4 or optional 2.6 kernels with ALSA
 support.  ALSA will be the standard in Etch.

 In GNOME:
 --------
 Right now, esound is pulled in by gnome-desktop-environment, but isn't
 started by default.  esound is known to be buggy, and I'm happy with
 the conservative choice of not enabling it as it permits applications
 using OSS or ALSA to just work.

 Some GNOME apps rely on GStreamer (Totem, Rhythmbox, Sound-Juicer)
 audio-output, others on esound, some even have ALSA and/or OSS plugins.

 In GStreamer:
 ------------
 GStreamer ships schema files which have the default audiosink set to
 OSS, new users will have their audiosink set to this.

 GStreamer applications outputting sound use a "gstreamer0.8-osssink |
 gstreamer0.8-audiosink" dependency.

 Future GStreamer releases might offer an autoaudiosink using whatever
 seems appropriate at run time (no configuration needed): esound if
 available, artsd if available, etc.


 2/ Proposed plan

 OSS should be supported, but the default is ALSA in the lower layers,
 and that should be reflected in the upper layers, that means that I
 propose:
 - shipping GStreamer with default audiosink alsasink, a NEWS.Debian
   inviting to change the current audiosink of existing users to
   alsasink if possible (this can be done via gconf-tool or via GNOME
   preferences)
 - requesting all GStreamer apps outputting sound to depend on
   "gstreamer0.8-alsasink | gstreamer0.8-audiosink" (Rhythmbox, Amarok,
   Totem, Sound-Juicer, ...) eventually with a NEWS.Debian suggesting
   the installation of gstreamer0.8-alsa
 - (the current configuration where esound is not started is fine)


 What do *you* think?


 I'd like some input from people using non-OSS/ALSA/esound audiosinks,
 eventually from people not using GNOME, so I might post this as a more
 general question.  In particular, I'm interested in knowing if any
 particular configuration changes are needed, and what packages need to
 be changed.


 Right now, this seems feasible in the GNOME + GStreamer world
 separately from the rest of the distribution, if you see any reason why
 the transition should happen all at once, I'm interested in bringing
 this up to debian-devel.

   Bye,

-- 
Loïc Minier <lool@dooz.org>



Reply to: