- disable stream_ReadLine debug.
* ALL:
- added a new demux2 module to handle Kasenna MediaBase
metadata files. These are application/x-sgimb masked
'mpeg' files on the web. They are in reality mostly constructors
for rtsp links. The reference to SGI is because the MediaBase
solutions originate from SGI.
VLM is a little media manager. It is designed to launch several souts
with a unique VLC. It is a kind of "server playlist". For now, it is
only usable with the little telnet interface I wrote. The features are:
- complete and barely coherent command line set for media setup and
control
- multiple input list by media (like a playlist)
- loop by media
- load/save config file (human readable)
- script implementation (for scheduling, for example)
If you want to test it, launch VLC with -I telnet, and type "help".
Beware of the type of the media: vod or broadcast. vod will be used
later.
TODO:
- rewrite the telnet interface (beurk)
- change error message system and create a http interface for VLM
- implement RTSP clients administration (with vod medias)
- write a complete documentation...
- Always have a space between ";then" the configure script does not like this. It should be "; then" or better put "then" on the next line.
- Add a AC_ARG_ENABLE for cdda
* modules/demux/dts.c: implemented DEMUX_GET_TIME and DEMUX_GET_LENGTH.
* modules/access/cdda.c: got rid of the cdda demux and add a wav header at the beginning of the data.
This allows to correctly support DTS audio cd (demux will detect what kind of CD it is).
* modules/demux/wav.c: code cleanup and fixed seeking.
to give it a try (QT is the default method).
+ Known issues: no fullscreen yet, memory leaks, ugly code, ...
+ Known goodies: it's a few % faster than QT here (playing a DVD). If
experienced OpenGL and Altivec programmers could have a look and
give some advice, I guess the performance increase might become
noticeable.
* modules/gui/skins2/src/var_manager.cpp: delete the variables in the
right order to avoid invalid reads in the destructor
* modules/gui/skins2/parser/xmlparser: skeleton of a new parser based
on the text reader API of libxml2
* New playlist import/export system:
-export plugin in misc/playlist (contains one module for each file type)
-open plugin = special demuxer2, that can be opened from the command
line or the graphical interface.
NB: the demuxer "playlist" is forced to earn speed, but this does not
work yet, as we have a demux2
* modules/gui/wxwindows: use the new import/export.
* src/playlist/group.c : fixed boundary check
* src/playlist/item-ext.c: correctly reset stuff when clearing the playlist
* modules/codec/ffmpeg/demux.c: new experimental demuxer using libavformat (most of the work done by fenrir).
This demuxer should allow us to support "frindge" formats (FLIC, SWF, etc...).
The demuxer has been put here because of its dependancy on libavformat and libavcodec.
* modules/codec/ffmpeg/ffmpeg.c: re-arranged the codecs list.
* modules/gui/skins2/Modules.am: renamed the plugin name into skins2 so it doesn't conflict with the old plugin.
* modules/gui/skins2/vars/playlist.cpp: fixed invalid cast.
this required new libvcdinfo API - so change to configure.ac for
pending new libcdio/vcdimager releases.
Additional libvcdinfo routines also means less VCD twidling inside vlc
code, thus simplifying what is already a bit complicated and likely to
get more so.
+ Fixed internal symbols for version numbers containing "+".
+ Added a --enable-caca flag.
* modules/video_output/caca.c:
+ Updated libcaca video output to the released version.
+ Added --caca-dithering flag (none, dither, random).
* more verbose header parsing code
* support for compressed announces if zlib is available (using some code
from Mosu from the matroska project)
* if the sdp packet contains more than one m= line, pass it to the
livedotcom module
configure.ac:
* link sap with zlib if available
modules/demux/livedotcom.c:
* accept sdp from the sap module
download a precompiled binary package, or to build all the packages
from the sources, at the user's option. It is currently written for
Mac OS X, but could easily be ported to other platforms.
+ modules/gui/beos/*: use hotkeys
+ include/vlc_keys.h, src/misc/configuration.c:
renamed modifiers into vlc_modifiers so it does not conflict with
BeOS' modifiers()
code from www.metakine.com.
CAUTION : the code is currently very ugly, with static variables, no
code convention, etc. I check it in "as is" because it may be useful
to some people.
Don't expect a bitrate reduction of more than 30 % with acceptable
quality.
* modules/demux/mpeg/ts.c: Useful debug message
things to write to implement competely the standard but this decoder
will work in most of the pictures subtitles cases. Text subtitles
decoding is not implemented at all.
This decoder has been validated with CAVENA encoders after few
remarks of their developpers.
include/*: addded a p_spuinfo field in the decider_fifo_t and the
es_descriptor_t
codecs.h: added a dvb_spuinfo_t structure to identify the id of the selected
track.
modules/demux/mpeg: added the support of the dvb subtitles track. Need last
libdvbpsi (0x59 descriptor) to compile. The demuxer
creates n fake ES as each ES carries up to 256 subtitles
tracks.
This work is part of the work done by Anevia for the "Gran Theatro del
Barcelona".
fixed a link issue on BeOS
+ modules/demux/util/sub.c:
gcc != 3 fix
+ modules/gui/beos/*:
compile fixes; never replace int32 with int32_t, these are not the
same on BeOS (long/int)
I have not kept the compatibility with older faad version (but check is
still be done in configure.ac). (I didn't like it as old version are
really buggy).
Isolated the video postprocessing routines in postprocess.c
* modules/codec/ffmpeg/encoder.c, modules/codec/vorbis.c, modules/stream_out/transcode.c,
include/vlc_block.h, include/vlc_codec.h: extracted the encoders out of transcode.c.
transcode now uses encoder plugins (currently ffmpeg, vorbis and theora).
PS: transcoding is currently a bit broken but I'll fix that ASAP.
* created a variable "key-pressed" in p_vlc that can be used to
report or read key presses
* allways try to start hotkeys interface in the background
configure.ac, modules/control/Modules.am, modules/control/hotkeys.c:
* a new control module that reads keypresses from "key-pressed" and
performs an action depending on what is asosiated with this key.
Also uses nice on screen messages to indicate what action was performed
modules/video_filter/filter_common.h:
* removed code to pass old style key reporting through filters
modules/video_output/x11/xcommon.c:
* report key-presses through the new mechanism. Incomplete, I need help
with this.
* all: changed the prototype of vout_AllocatePicture() and vout_InitPicture().
* src/video_output/vout_pictures.c: new vout_InitFormat() function that initialises a video_frame_format_t structure.
* include/vlc_video.h: extended video_frame_format_t.
* modules/codec/rawvideo.c: ported to the new decoder api.
* modules/video_output/x11/*: simplifications.
+ Detect <mozilla-config.h> on newer Mozilla versions.
* mozilla/vlcpeer.cpp:
+ If neither NP_UNIX, NP_MACOSX nor NP_WIN are defined, default to NP_UNIX.
+ Use <mozilla-config.h> whenever possible.
* modules/demux/ogg.c: added support for speex to the Ogg demuxer.
* modules/mux/ogg.c: added support for speex to the Ogg muxer.
* src/misc/variables.c: when creating choices object variables with inheritance enabled, automatically add the inherited value to the list of choices.
* modules/access/udp.c: use an object variable and inheritance for "udp-caching".
* src/input/input.c: disable the hack for filenames with colons on win32 because filenames can't contain colons.
+ updated libebml/libmatroska detection as matroska headers no longer
include ebml headers directly. Hope you upgraded it - if not, you'll
have to set your CFLAGS accordingly.
+ on BeOS, libmkv_plugin.so has to be linked to libstdc++
modules/control/http.c:
+ fix for gcc != 3
+ First try to detect the OS X native dynamic linker before falling
back to dlfcn.h.
+ Enhanced the dlopen-style loader detection.
+ Added a check for HP-UX's shl_* functins.
* src/misc/modules_plugin.h.in:
+ Support for the dylib dynamic loader on OS X.
+ Support for the shl dynamic loader on HP-UX.
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.
The final aim of this new api is to make it possible to use the decoders from other modules like the transcoder for instance.
Only a few decoders have been ported to the new api (a52, libmpeg2, dts, vorbis, theora) so the old api is still supported.
Don't hold your breath, there is still much work to do before we reach this goal.
* modules/codec/a52.c, libmpeg2., dts.c, vorbis.c, theora.c:
Converted to the new api.
Merged the a52 and vorbis packetizers in their respective decoders (removes a lot of code duplication).
New dts and theora packetizers (merged in their respective decoders).
Much work still needs to be done, like audio support, adding plenty of configuration options, etc... But the video part is already working quite well here.
special goom tree source as I was unable to use standard goom library.
(I will provide it once mmx/ppc included)
* modules/visualization/Modules.am: added goom
* include/vlc_block.h src/misc/block.c: introduce a new data block
api (not yet tested, ported from my local new input work).
in use by the OS. It's pretty nifty. if available it uses nl_langinfo. Then
it falls back to trying setlocale (exept on OSX, where this funtion is useless)
and then falls back to LC_ALL, LC_TYPE and LANG environment variables.
If only the LANG variable is available (like on osx) it tries a countrycode
to charset mapping (making an educated guess ).
On windows it retrieves the charset with GetACP().
It also has an aliases system to transform incompatible charset description
strings to libiconv compatible charset descriptions. If you want you can even
retrieve the charset on OS/2.
* modules/codec/subsdec/subsdec.c: By default we try to open a subtitle file in
the systems charset (or language mapped to charset). It's no more than a guess,
and if you open Latin1 subs on your Japanese system you will still need to
specify the encoding by hand, but it's better then nothing.
* src/playlist/playlist.c: spelling error in a comment.
* modules/visualisation/visual/* :
visualization system. It supports multi-effects.
At the moment:
-dummy (guess what ? it does nothing)
-scope (adapted and fixed from sam's one)
-random (displays random plots)
Todo:
- spectrum analyser
- movement effects (rotations, scrolls,...)
- blur effect ?
Andre Pang