| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
|
|
This way external plugins can use their own gettext catalog for i18n.
|
|
|
|
This is the start of a new experimental branch, with the first objective being
the replacement of the get_description and get_identifier functions with
direct-access strings.
The reason for this change is to reduce code size and time of execution.
By replacing the functions with direct-access strings there is one less call
to be done in those cases where the description has to be fetched.
The solution is not yet definitive though, there are a couple of problems to
take care of:
- the use of N_() still makes it easy to internationalise the strings, but it
requires for the string to be found on libxine2 catalog, which is not
exactly a nice solution for external plugins;
- it would be simpler to re-use the id field in plugin_info_t, and then move
description there; it should reduce memory usage for the class structures;
- I'm not really aware of any reason why get_description and get_identifier
were used beside the idea of making i18n simpler.
This probably would break a couple of frontends, especially if they have some
internal plugins (like post-plugins), so it needs to be reviewed carefully
before merging in 1.2 branch. My current goal is to get this in before 1.2
though, rather than waiting for 1.3.
|
|
|
|
there.
Instead of using a thread to watch for completion, use mutexes where needed.
Don't wait for operation completion for volume changes, they can happen asynchronously too.
|
|
|
|
Instead a condition is signalled when adding new entries, and waited for if no entries are present.
|
|
no more race conditions.
|
|
|
|
Note that the code is incomplete: the infinite loop without sleep will create unneeded wakeups and waste resources.
|
|
|
|
|
|
|
|
--HG--
extra : transplant_source : %15%CD%0B%2C%F2g%03%B9%25%D4%B2%8BQ%1F%EB%BB%8F%28%9B%C8
|
|
- Allow visible region updates.
- Resize OSD at drawing time.
--HG--
extra : transplant_source : %17%ECZ%D0j%2C%BC%06%7CO%3A%F0%C7u%A3%86%DE%8Bb%D3
|
|
|
|
|
|
Compilation was "broken" by ffmpeg rev. 9283.
|
|
--HG--
rename : src/demuxers/demux_ogg.c => src/combined/xine_ogg_demuxer.c
|
|
handle that case
|
|
In ao_loop only read the first buffer and remove it from the fifo only when the
buffer has been written or is about to be discarded. This fixes the race between
ao_loop and ao_close for good. Now fifo_remove_int may signal empty again right
after removing the last buffer from the fifo.
|
|
|
|
but only when a read was requested and the fifo is empty.
Rationale: ao_close uses fifo_wait_empty to make sure all buffers were written
to the driver before it calls close on the driver. But if empty is already
signaled when ao_loop just reads the next buffer then ao_close might close the
driver before ao_loop has a chance to send the buffer to the driver.
|
|
buf must be != NULL because of the while(!fifo->first) before and the buf->next after
|
|
not played yet.
When closing the pcm device make sure that everything that was written to the
device actually gets played. This was the race we were seeing with Ogg Vorbis
and wav playback where a delayed driver->close would fix playback. Notice that
blocking mode needs to be used for snd_pcm_drain, otherwise the call would be
a noop.
|
|
|
|
|
|
|
|
|
|
Some fixups were required for the kfreebsd patch; needs testing.
--HG--
rename : src/video_out/libdha/ports.c => contrib/libdha/ports.c
rename : src/demuxers/demux_ogg.c => src/combined/xine_ogg_demuxer.c
rename : src/libxineadec/xine_vorbis_decoder.c => src/combined/xine_vorbis_decoder.c
|
|
Negative values would be accepted, resulting in junk in XINE_META_INFO_GENRE
or a segfault.
|
|
|
|
|
|
|
|
|
|
This way you can get a safer build for binary redistribution.
|
|
|
|
more accurate (compared to oggdec); fix one debug printf
|
|
|
|
DEMUX_FINISHED from send_header
|
|
|
|
|
|
|
|
|
|
It changes some "defined (__FreeBSD__)" into "defined (__FreeBSD_kernel__)"
and "__FreeBSD_version" into "__FreeBSD_kernel_version".
The changes are performed on places, where feature of kernel is tested. On
some other places, feature of userland/libc is tested, on them "defined
(__FreeBSD__)" remains.
As proposed, include/configure.h cause __FreeBSD_kernel__ be defined if
__FreeBSD__ is defined.
--HG--
extra : transplant_source : %25%96K%05%E4Y%B15%94%60%15%FE1%8Ah%26Xy%8C/
|
|
|
|
|
