Add new file timeline/tl_matroska.c. Move the code that parses
ordered chapter information from Matroska files and creates the
timeline structure based on that to the new file.
Initialize the format parameter given to open_stream() in the moved
code. The previous uninitialized value shouldn't have caused any
visible effects.
Drop support for specifying demuxer types by numeric ID (options
-demuxer, -audio-demuxer and -sub-demuxer). Stop printing the numeric
values in "-demuxer help" output. Convert the list of DEMUXER_TYPE_XXX
defines to "enum demuxer_type".
Remove some unused lines from demuxer.h. Make some demuxer.c functions
static. Move new_ds_stream() declaration from demuxer.h to stream.h
(the function is defined in stream.c). Clean up some code in mplayer.c
that had commented-out free_demuxer_stream() calls.
Force Makefile to always run version.sh to potentially regenerate
version.h. Drop compiler version and 'git-' prefix from version
number. Match only git tags starting 'v'+number when generating
version number; leave the 'v' out from the result.
Change DVB SPU stream format in TS demuxer so it can be decoded by
libavcodec (as soon as lavc is fixed not to fail just because of an
extra padding byte).
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@32866 b3059339-0415-0410-9bf9-f77b7e298cf2
Fix r32587: the previous approach to return subtitles in time broke
DVB subtitles due to returning incomplete packets and even for
PGS subtitles resulted in incorrect pts values for the sub packets.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@32864 b3059339-0415-0410-9bf9-f77b7e298cf2
Do not pointlessly cast the return value of memalign().
memalign() returns void*, which is compatible with any pointer in C.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@32850 b3059339-0415-0410-9bf9-f77b7e298cf2
add apco and ap4h fourcc to prores decoder
add ai55 and ai15 fourcc to h264 docoders
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@32843 b3059339-0415-0410-9bf9-f77b7e298cf2
Use color type getter instead of direct access to private member.
Using the getter is mandatory since recent libpng 1.5 release.
Patch by Gianluigi Tiesi (mplayer - netfarm it)
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@32840 b3059339-0415-0410-9bf9-f77b7e298cf2
Escape character values >= 127 in URLs as required by the RFC.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@32833 b3059339-0415-0410-9bf9-f77b7e298cf2
Check that rlen is valid before using it to increment a pointer.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@32832 b3059339-0415-0410-9bf9-f77b7e298cf2
Disable libmpcdec (musepack) support by default, FFmpeg has a decoder
that works better for us and is used by default.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@32826 b3059339-0415-0410-9bf9-f77b7e298cf2
Convert cache_fill_status into a function so we always get the latest
state, not whatever it was after the last read.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@32818 b3059339-0415-0410-9bf9-f77b7e298cf2
Update PAUSED status line with cache fill status if it changed.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@32819 b3059339-0415-0410-9bf9-f77b7e298cf2
configure had a check for some libraries specific to the maemo
platform. Apparently the code which would have used the results of
this check was never added. Remove the unused check.
Previously the default with --enable-translation was to install all
language files if --language-msg was not explicitly specified. Change
the default to the languages set with the --language option or LINGUAS
environment variable if those are set; if neither is set then the
default is still to install all message languages.
The XML documentation was badly outdated, with various obsolete,
useless and and sometimes actively harmful advice. Delete a lot of the
obsolete stuff, without spending much effort to replace it for now.
In the FAQ section I removed some questions completely; in other cases
I only removed the answer if I thought the entry might be worth
keeping but the answer would need to be updated.
Abort if FFmpeg libraries are not found unless --disable-ffmpeg was
explicitly specified. Note that even if you disable FFmpeg, libavutil
and libswscale are still required and you'll need to add linker flags
for them manually; I didn't try to clean up that case (few people
will want to do that).
Also fix the fallback test used if pkg-config doesn't find the
libraries: fix wrong header name and drop forced -lavcore which isn't
present in all FFmpeg versions. This shouldn't make a difference for
normal use since the pkg-config test should succeed if the libraries
are available.
af_lavcac3enc: use old SampleFormat names without AV_ prefix, the
latter were only added in 2010-11
vd_ffmpeg: add ifdef around CODEC_ID_LAGARITH use
demux_real: use ffmpeg_files/intreadwrite.h
stream/http.c, stream/realrtsp/real.c: define AV_BASE64_SIZE macro for
old libavutil versions lacking it
Enter and Tab are the only named keys (rather than identified by the
printable character they produce) with code below 256. Add a special
case to recognize the Shift modifier with them.
The keycodes.h file contains a KEY_CTRL define, then various control
keys expressed as "KEY_CTRL + 0" and so on. Back in 2002 when the key
name table in input.c was created this KEY_CTRL define was mistakenly
interpreted as a key name, apparently confusing it with the Ctrl key
(even though the input system didn't handle that key back then). As a
result there was an incorrect key name entry with the same key code as
backspace and name "CTRL". This incorrect entry was used when printing
the name of the key. Delete it.
vo_x11_create_vo_window() only called vo_x11_update_geometry() if no
window position had been specified by -geometry or -xineramascreen, to
avoid overwriting the specified position with values from the existing
window. However window size should be initialized to the existing
window here, and setting new window title for -use-filename-title is
also done in vo_x11_update_geometry() (for whatever reason, it doesn't
match what else that function does). Change the code in
vo_x11_create_vo_window() to always call vo_x11_update_geometry() for
size variable and window title updates, but add a flag that tells it
not to update position variables.
The libavcodec AC-3 encoder was changed to use floats, and take
floating point samples as input (the fixed-point version is still
available under the new name "ac3_fixed"). This broke af_lavcac3enc
because it blindly assumed without checking that the "ac3" encoder
would take signed 16-bit integer samples. Improve af_lavcac3enc so
that it checks the sample formats supported by the encoder and can
handle either int16_t or float.
Perhaps an option to keep using integer input but instead switch the
encoder name to "ac3_fixed" for new libavcodec versions would have
some value. Then again, maybe not. Using the preferred data format of
the default "ac3" encoder should normally be best, so probably better
not add such an option unless real need appears.
One of two alternative code parts passing codec extradata to
libavcodec didn't add the buffer padding that libavcodec requires,
resulting in invalid reads beoynd allocated memory area. Fix.
libvorbis and libvorbisidec cannot both be enabled at the same time.
Change configure to enable libvorbis by default if both are
available. It's the more common library, and if someone want to do a
special build for FPU-less systems he can select the library
manually.
Drop internal copy of the tremor library. Note that the internal ogg
demuxer (which is still sometimes useful to work around libavformat
ogg demuxer problems, though it's itself quite buggy) now cannot be
compiled without either external libvorbis or libvorbisidec (tremor).
The "Core dumped ;)" message printed after finishing a stream dump is
known to confuse users but was kept as "humor". Change it to say
"Stream dump complete." instead.
About the only useful part of the BUGS section on the manpage was a
link to mplayerhq.hu, and that doesn't really apply any more (includes
references to Subversion etc). Remove that section and update/remove
some outdated info.
The resulting semantics of this flag are weird enough that they're
unlikely to be what is wanted in any situation. Remove the flag and
convert the two options using it, -screenw and -screenh, to use
CONF_NOSAVE instead. I'm not sure why those specific options had the
flag and if any flag is really needed, but I don't want to check in
detail now and using CONF_NOSAVE should keep about the same behavior
in practice.
A bit more detail about the weird behavior this flag had:
When not using file groups, the flag had the same behavior as
CONF_NOSAVE, namely that when switching files the option would not be
reset to the global value (only possible file-specific settings were
applied). When using file groups, group-specific options would apply
to the _first two_ files in the group, but for the rest after the
first two, settings would not be reset when changing files (wtf?).
This was a result of the following sequence:
1) push higher-level settings, enter group
2) apply group-specific settings
3) push settings before applying ones specific to file 1 in group
4) apply file 1 settings, play file 1
5) pop settings to return to group settings
6) push settings before applying ones specific to file 2
7) apply file 2 settings
8) pop settings
Here the option was set at 2). 3) saved it because it had been set
after last push, so 5) restored the setting and it was used for file 2
too. However 6) no longer saved it because there had been pushes after
the original setting in 2), thus 8) no longer restored the setting and
the option was no longer forced to any particular value when playing
further files after that.
Most of the time users want to keep current fullscreen state when
switching to another file, rather than always resetting to the global
default that was determined when the command line was parsed. Add the
CONF_NOSAVE flag to the -fs option to achieve this. The change means
that the fullscreen state from the previous file is kept unless there
is an explicit file-specific -fs or -nofs setting for the new file
(or in case file groups are used, moving to the new file enters a
group with explicitly specified state).
vo_zr was the last user of the "func_full" option type. I think it's
better if future features are implemented using more straightforward
option functionality. Delete the func_full implementation.
There were multiple files specific to Zoran support, and they also
depended on internal FFmpeg headers (so it would probably have been
hard to get them to compile now even if you tried). It's obsolete now,
so just drop the whole mess.
dxr2 support had been broken quite a while and nobody noticed. There
were finally commits to fix it in the svn repo, but rather than apply
those I'll just drop dxr2 support.
By now VIDIX is too obscure to justify the amount of code and
complexity it requires in the sources. Although there is no pressing
need to drop it just now from a code point of view, I'll rather remove
it before release than release with VIDIX support and then drop it
later.
Some of the manpage mentions of VIDIX were in "this option supported
for these VOs" lists that looked outdated and failed to mention vdpau
for example. Replace such incorrect lists with a generic "not
supported for all VOs" mention.
Uoti Urpala
cehoyos
reimar
faust3
diego
compn
cboesch
michael