gst-plugins-good0.10 package

David I. Lehn dlehn at
Tue Dec 20 01:33:30 UTC 2005

* Sebastien Bacher <seb128 at> [2005-12-19T18:39:39-0500]:
> * the different sinks:
> gstreamer0.10-aa, gstreamer0.10-auto, gstreamer0.10-caca,
> gstreamer0.10-esd, gstreamer0.10-oss 
> * the misc package:
> gstreamer0.10-misc-good
> The Depends are:
> Depends: gconf2 (>= 2.12.1-4ubuntu1), libavc1394-0 (>= 0.5.0), libc6 (>=
> 2.3.4-1), libcairo2 (>= 1.0.2-2), libdv4, libflac7, libgconf2-4 (>=
> 2.11.1), libglib2.0-0 (>= 2.9.1), libgstreamer-plugins-base0.10-0 (>=
> 0.10.0), libgstreamer0.10-0 (>= 0.10.0), libjpeg62, liboil0.3 (>=
> 0.3.3), liborbit2 (>= 1:2.10.0), libpng12-0 (>= 1.2.8rel), libraw1394-5,
> libshout3, libspeex1, libvorbis0a (>= 1.1.2), libxml2 (>= 2.6.22),
> zlib1g (>= 1:1.2.1)
> Any comment on the current split? Should we move some of -misc? 
> The current Depends are quite reasonable. Is there some people wanting
> to run applications without this package? and
> the .schemas are shipped with it and required by the applications, if
> that's the case we may want to create a -gconf too (according to the
> misc requirements I'm fine with both way and that's probably fine for
> other people as well)

Back when I first did the -misc seperation the dependencies were just
simple things that ~99% desktop users would have installed already.  It
kind of balloned up to a bigger list because as gstreamer plugin scope
increased it seemed silly to create a new deb for each dependency.  I
had always hoped to revist the problem and have a better solution
magically appear.

Are you making a top level -plugins-good package that depends on
everything else?  Should there be a core -plugins package that depends
on the -plugins-good, -plugins-bad, -plugins-ugly, -ffmpeg, etc
packages?  Ie, something to install all of gstreamer.

Why should sinks get special packaging and not sources or codecs?

>From an end users perspective the delta pain of the few sink plugin deps
above is probably not that bad.  Why not just throw them all into

I've always wondered if the kitchen sink approach is really going to
bother anyone.  I never had any data from my earlier work if anyone
really cared that it was split up.  I always imagined 99% of the users
would install everything via the top level -plugins package.  But I
really don't know.  People don't seem to complain too much about all the
crap GNOME and KDE desktops pull in.

Cons of uber package:
- Bloat
- Lots of deps, may not need many of them
- Bloat
- Must deflect users complaining about deps
- One broken lib can cause whole package to be uninstallable

Pros of uber package:
- Disks bigger and networks faster these days. (kind of lame argument)
- Packaging is _much_ easier.  I burned out on it.
- Users will have everything installed. "Batteries Included"
- Easier for gstreamer apps to depend on a few packages rather than
  every codec, sink, source, etc they may want to use.


More information about the Pkg-gstreamer-maintainers mailing list