added wma1/2 decoder... :))

 * libioRIFF : now useless files.

 * finally added languages/subtitles selections menus, found and got fixed two
 bugs in kdelibs in the process, but only one of them will get into kde 3.1.
 * made the location of the ui.rc file configureable, usefull for testing.

faad/decoder.c : minor compile fix

* modules/demux/ogg.c: we now set p_input->stream.i_mux_rate.

* ALL: removed GetPES and NextPES, we now use input_ExtractPES everywhere instead
   of these. The bitstream facility has also been changed to use input_ExtractPES
   and now stores the current PES in its structure.
   Introduced input_FlushPESFifo() and CloseBitstream().

some regression.

 * all : fix for new use of p_selected_area->i_tell.

 Just a word: now we cannot update anymore p_input->p_current_data
without updating i_tell. I known that it's used for skipping some bytes in
mp3 demux (not yet fixed), I don't known if it's used elsewhere....

* ALL: p_selected_area->i_tell is now update inside input_SplitBuffer instead
   of inside the pf_read functions of access plugins. That means i_tell now
   actually reflects the position of what is being decoded and furthermore it
   is updated more often.

    not 5, to avoid using a color used by another application.
  * ./modules/video_output/directx/events.c: support for drawing in a parent
    window instead of creating our own; still a bit flakey.

  * ./mozilla/vlcshell.cpp: we can now compile the Mozilla plugin so that it
    does not call libvlc (for testing purposes).

* include/aout_internal.h: oops, forgot to commit this in my last changes.

* modules/audio_output/oss.c: when now set the fragment size of the OSS device
   according to the size of an audio sample.
   For buggy OSS drivers, we now take into account the real size of the OSS
   internal buffer when we check how much this buffer has been filled.

    goal is to provide dependencies for shared libraries.

    int16_t) instead of our custom types (s8, u8, s16) and you are encouraged
    to use these types from now. Of course the old types are still available
    for backward compatibility.

* modules/audio_output/oss.c: Finally fixed! There also is a new config option
   (--oss-buggy) to work-around buggy oss drivers.

 * input_ext-dec.c : add a call to GetPES in NextPES when fifo is empty.

* src/input/input_ext-dec.c, include/input_ext-dec.h, modules/codec/araw.c:
   renamed input_NextPES into input_ExtractPES and factorised its code.
* modules/codec/ffmpeg/ffmpeg.c, modules/codec/faad/decoder.c,
   modules/codec/cinepak/cinepak.c: make use of NextPES and GetPES from
   src/input/input_ext-dec.c.
* modules/codec/a52.c: fixed uninitialized variable.

* ALL: I did add these files, but forgot to commit them ( ouarf ouarf :)

* ALL: New ogg demux and vorbis codec modules for preliminary support of
   ogg/vorbis streams.

* ALL: fixed a bunch of memory leaks.

* Miscellaneous small fixes.

    For the moment it spawns a separate window and seems to behave nastily
    (read: crashes) when the target has sound, but the javascript interface
    works perfectly, for instance.

    To build it, I installed Mozilla development files in my mingw32 dir
    and hacked mozilla-config so that it uses these files. I also hacked
    jri_md.h and jni_md.h to put "defined(_MSC_VER) || defined(__MINGW32__)"
    where it read "defined(_MSC_VER)". No other changes were necessary.

  messages,
* Updated BeOS and QNX outputs with respect to i_channels policy.

(input_NextPES);
 * araw.c : use this new function.

 * modules/demux/asf/*: some clean up

usable, but it begin to give good result.

* include/vlc_threads.h: Avoid locking the interface,
* ALL: reworked all audio output plug-ins to correctly take into account
  i_channels. At present this is known to break OSS, probably others.

    the DirectX plugin (please test compilation).

  * ./modules/video_output/x11/xcommon.c: we transmit the click only at
    release time.

  * ./modules/access/dvdplay/intf.c: instead of probing the mouse on each
    loop iteration, we set a callback on the corresponding variables.

    can now use var_* functions from within your callback, they can even
    trigger callback themselves, to any level. The main restriction is that
    you shouldn't meddle with a variable that is already being triggered by
    the current thread (other threads will just wait).
  * ./src/misc/objects.c: fixed a deadlock in the "tree" command.
  * ./modules/misc/testsuite/test4.c: added a "callback-test" command to the
    rc interface to test callback loop detection and concurrent triggers.