* Moved modules_builtins.h.in into src/misc. Please remove
    include/modules_builtins.h in your local tree.

  * DirectX plugin by Gildas Bazin <gbazin@netcourrier.com>.

    Notes: o I called it 'directx' instead of 'windx' because it
             made a lot more sense to me.
           o Completely untested. I didn't manage to compile it yet.
           o Code in vout_directx.c will go through a big rewrite when
             vout 4 is finished. Xav is already working on this.

*fixed a lock issue in the interface introduced in my last commit

  * Win32 plugin support by Gildas Bazin <gbazin@netcourrier.com>.

  * BeOS fixes. renamed iovec.h to input_iovec.h because of namespace issues.

  * miscellaneous Win32 fixes by Gildas Bazin <gbazin@netcourrier.com>, sorry
    for the 2 weeks delay.

  * Win2000 DVD input by Jon Lech Johansen <jon-vl@nanocrew.net>.

  * Applied patch from Jon Lech Johansen <jon-vl@nanocrew.net> to compile
    vlc with MS and Intel C/C++ compilers.

-Fixed some bugs in interface (language menus in ts, misnamed fields)

-Play/Stop work in network mode to start/stop decoding of the stream

I have a segfault in imdct_3dn(93). But I can't underdand assembly.

  * Fixed a bug introduced in my previous commit which broke DVD input.

  * Hrmpf. Forgot the most important file.

  * Fixed the BeOS compile typo.
  * The ALSA plugin is back.
  * A gift for Gildas: plugins don't rely on backlinking any more. The
    drawback is the dirty magic in include/modules_export.h which has to
    be included whenever a module needs one of the functions in it. But
     a) we should less and less rely on those functions
     b) no plugin code modification was needed

   The BeOS and MacOS X modules may be broken when compiled as plugins,
   I'll check this later when I can reboot under BeOS.

-Bugfix in the interface to access the network specific features.

-Some base of synchro in ac3 spdif

  * Added a wrapper for readv() on platforms which don't support it. The
    network support now compiles under Win32, but still doesn't work, the
    select in plugins/mpeg/input_ts.c never returns any data.

    Is there anything like strace(1) under Win32 to debug this ?

It compiles under Linux and Win32 (yeeaaahaaa, haven't broken anything
:-))
(added the WinSock library initialization and cleanup)
Could someone test if it does really function ?

  * Of course my Win32 patch broke the Un*x version.

  * Forgot a little #ifdef, sorry.

  * Win32 compilation fixes, esp. the MAC retrieval.

  * Fixed my latest BeOS support breakage.

  * Fixed a warning for *BSD compilation.
  * Fixed a typo in yesterday's Darwin ioctl fixes.
  * Patch by Eugenio so that the OS X interface and video output handle the
    aout and vout banks.

  * Forgot a file for BeOS.

  OS X fixes:

  * Applied Eugenio's fix to the DVDioctl driver. I didn't test it, but
    it should not hang the kernel anymore.

  * Ported most dvd_ioctl.c functions to the DVDioctl driver: ReportAgid,
    ReportChallenge, ReportKey1, ReportASF, InvalidateAgid, SendChallenge,
    SendKey2. They should comply with the "Mt. Fuji Commands for Multimedia
    Devices" paper.

    Last remaining problem: ReadCopyright and ReadKey. The version of
 Darwin I have here only has reportKey() and sendKey() functions in
 the IODVDBlockStorageDevice class (pages 413 and 449 of the Mt. Fuji
 paper), but we also need the READ_STRUCTURE (page 357) which I couldn't
 find anywhere in the kernel headers. Either I missed it, or we'll have
 to wait for Apple to implement it (or maybe we can emulate this command
 with some other IOKit commands).

Specific files are glade generated files and intf_* files. All the
remainders are common to gtk and gnome and contain all the callbacks for
both interfaces.

title/chapter/audio/spu/angle, we tell the input to do intead of doing
it inside the interface. It results in fewer locks during the changes.

*The dvd plugin reads again blocks of 32 sectors to gain speed since
there are no more lock-ups during stream change. I've also created a new
file that contain function to described video and audio streams with ifo
datas.

The changes are made only for gtk. We should update the other interfaces
soon (beos at least).

There is a memory corruption somewhere that make the vlc crash. I've no
idea where to search.

* Remove kmudge for ac3 on MacOS X

  * We now do pthread detection for FreeBSD like the XMMS guys do. It
    works even better.

  * FreeBSD pthread fixes.

  * AC3 IMDCT and downmix functions are now in plugins, --imdct and
    --downmix options added.

*added a b_stopped flag to interface playlist to have a stop function in
interface plugins. As a consequence, the playlist no longer loops on end
but wait for the user to press 'play' or to add another item (gtk/gnome
interface). I think that other plugins are broken but it shouldn't be
difficult to repair: just add a 'p_main->p_playlist->b_stopped = 0' in
play functions and 'p_main->p_playlist->b_stopped = 1' when you have to
stop.

*Updated gtk interface so that it has the same features as gnome. I
think I will merge the code so that we don't have to copy and paste each
change. Only the .glade file would be different.

If you have a PIII or a Athlon and you want to try this, just comment #if 0
in ac3_downmix.c and ac3_imdct.c and add in AC3_DECODER section of Makefile :
	src/ac3_decoder/ac3_imdct_sse.o \
	src/ac3_decoder/ac3_srfft_sse.o \
	src/ac3_decoder/ac3_downmix_sse.o \
	src/ac3_decoder/ac3_downmix_3dn.o \

Put a little note in the doc about "dvd devices"

  * Fixed a deadlock in spudec.

(jwbaker@acm.org)

Please notice that even if if builds with alsa 0.9beta, it's hardly
usable as it goes very often un Xrun mode. Though, it works fine with
current alsa CVS.