* modules/gui/wxwindows/*: got rid of the close button in the fileinfo and playlist dialogs.

* modules/gui/wxwindows/*: we now have only 1 global "Advanced Options" checkbox in the preferences panel instead of 1 for each section (KISS).

* modules/access/dshow/dshow.cpp: compilation fix related to ssize_t.
* share/vlc_win32_rc.rc: stringify version number.
* mozilla/*: compilation fix and cosmetic changes to the .rc.

* modules/stream_out/duplicate.c: fixed array overflow.

* mozilla/vlcshell.cpp: on win32, try to find the location of the vlc plugins from the registry key.

  - removed the defines for i_mode concerning random and loop. We use config
    variables for all of these.
* src/libvlc.h:
  - removed the enqueue playlist config option. It's no longer used.
  - added a repeat option. vlc will keep playing the same playlist_item,
    until the option is unset.
* src/playlist/playlist.c:
  - added a repeat, random and loop variable.
  - wxwindows should use these as well I think.
* modules/gui/macosx:
  - added the INTF_ABOUT_MSG to the about dialog.
  - implemented the about and repeat items.

* src/input/input_programs.c: p_input->stream.p_selected_program needs to be reset in input_DelProgram().
* modules/demux/mpeg/ts.c: use dvbpsi_DetachPMT() when a program is deselected.

* configure.ac, mozilla/Makefile.am: more win32 mozilla plugin build fixes.

* configure.ac, mozilla/*: the win32 mozilla plugin compiles again (most likely won't yet work).
 Added a --with-mozilla-sdk-path option (win32 only) to ./configure so you can pass the path of your mozilla sdk tree.
 You also need to set the XPIDL env var to the the path of the xpidl mozilla compiler when ./configuring.

 (Untested)

  - implemented the new input variables for control.
  - reevaltuated the locking mechanisms in the osx intf.
    a lot of this can now be removed, because of the new input structures,
    and the vout garbage collector of playlist.

    + "==" is not valid shell test, use "=" instead.
    + Set LANG for consistency of sort output.
  (NB: this really belongs to the toolbox script).

* po/pt_BR.po: Updated by Andr� de Barros Martins Ribeiro <andrerib at_ajato.com.br>
* src/libvlc.h: Added es (Spanish) to the list of choices.

  ( Yes it's backwards compatible ).

        use vout_Request to release the vout (so vout isn't closed when
        changing playlist item).

  recognized by VLC
* modules/code/quicktime.c: Added samr (AMR) to the quicktime decoder.
  Although it runs on OS X, it only plays back silence for some reason.
  Might be a problem with one of the converters?
* modules/gui/macosx/*:
  - 'fixed' several compiler warnings
  - removed some old no longer necesarry stuff

 * mp4: send complete atom for amr to the decoder.

 first search  the root of  the object  using p_parent and  not directly
 using p_vlc. If this first search  failed then we search using p_vlc if
 it wasn't the  case. (I'm not sure  it's harmless but I  think it's the
 correct behavour for FIND_ANYWHERE)

 * src/video_output/video_output.c  : we could now  use FIND_ANYWHERE to
 catch the  playlist (even when the  vlc is exiting. (Hehe, now visual
 effects won't reopen the vout every playlist item.)

New dvb and old satellite modules now use the *correct* order of unit for tuning to a satellite channel. All information must be supplied in KHz not in MHz !!!

Frequency et al is really specified in KHz and not in Hz. I got an order of magnitude wrong. The same mistake is in the old satellite plugin's and user interface for specifying the frequency. It is specified in MHz instead of the KHz it should be.

    listed in POTFILES.in

    deleted file encoder.c

    issue on Debian s390.

* modules/access/slp.c : do not add udp:@ automatically

* modules/stream_out/announce.c : always use the same service name
				  add udp:@ (hack !)

 * if the given vout thread is a filter, then don't sleep untill the display
date. This reduces the chance of pictures being late when reaching the vout
if one or more filters are in use
 * don't include current render time in the render_time sliding mean if it
is way too large ( as inspired by Simon Gittins ).
 * weight the current sliding mean tree times more than the current render
time when calculating the new render time. This should make the code more
robust to jitter in render time