Age | Commit message (Collapse) | Author |
|
in the future (it would delay the playback of a new stream).
CVS patchset: 3642
CVS date: 2002/12/23 00:51:31
|
|
- move frame discarding to inside the output loopss. it is much safer to
manipulate fifos from there avoiding any races.
- do not reject audio frames because vpts is smaller than previous. it
might happen on seeking but it's ok, because the sound driver may have
being flushed too.
- fix xine_demux_flush_engine so it doesn't call output flush too early
- remove scr adjusting
please report any problems...
CVS patchset: 3639
CVS date: 2002/12/22 23:30:29
|
|
new buffer methods to avoid deadlocks doing flush from video_out
CVS patchset: 3629
CVS date: 2002/12/22 15:02:06
|
|
CVS patchset: 3620
CVS date: 2002/12/21 19:23:01
|
|
CVS patchset: 3619
CVS date: 2002/12/21 19:20:09
|
|
CVS patchset: 3617
CVS date: 2002/12/21 18:39:35
|
|
just couldn't see it! :)
CVS patchset: 3616
CVS date: 2002/12/21 18:17:33
|
|
CVS patchset: 3613
CVS date: 2002/12/21 16:25:31
|
|
prevents the gui from drawing slider at wrong positions during seek.
CVS patchset: 3609
CVS date: 2002/12/21 16:13:43
|
|
CVS patchset: 3608
CVS date: 2002/12/21 16:12:25
|
|
different sizeof than integer.
- add extra_info structure to pass informations from input/demuxers down
to the output frame. this can be used, for example, to pass the frame
number of a frame (when known by decoder). also, immediate benefict is
that we now have a slider which really shows the current position of
the playing stream. new fields can be added to extra_info keeping
binary compatibility
- bumpy everybody's api versions
CVS patchset: 3603
CVS date: 2002/12/21 12:56:44
|
|
it's just a hack. Oh well, set it straight
CVS patchset: 3600
CVS date: 2002/12/21 03:03:15
|
|
'q' key).
When detecting a 'stall' condition in video_out, the 'flush()' on the video
decoder deadlocks with with the video_out close from the video_decoder
thread when video_decoder sees a BUF_CONTROL_QUIT message. The two threads
grab the streams_lock and the video_decoder.lock in reverse order, leading to
deadlock (deadlock happens quite often on solaris).
CVS patchset: 3596
CVS date: 2002/12/20 18:14:34
|
|
CVS patchset: 3590
CVS date: 2002/12/19 23:41:45
|
|
CVS patchset: 3589
CVS date: 2002/12/19 21:40:02
|
|
CVS patchset: 3579
CVS date: 2002/12/18 03:59:10
|
|
CVS patchset: 3558
CVS date: 2002/12/16 18:57:49
|
|
CVS patchset: 3557
CVS date: 2002/12/16 01:50:52
|
|
CVS patchset: 3556
CVS date: 2002/12/16 01:34:08
|
|
CVS patchset: 3538
CVS date: 2002/12/15 16:01:11
|
|
CVS patchset: 3521
CVS date: 2002/12/14 18:34:57
|
|
Attempting to initialize an already initialized condition
variable results in undefined behavior.
fifo->not_empty was initialized in fifo_new(); there should be no
need to re-initialize it in fifo_remove_int()..
CVS patchset: 3519
CVS date: 2002/12/14 16:33:59
|
|
ao_fill_gap() locks the driver, too ==> deadlock
CVS patchset: 3518
CVS date: 2002/12/14 16:17:57
|
|
CVS patchset: 3514
CVS date: 2002/12/13 21:31:38
|
|
CVS patchset: 3511
CVS date: 2002/12/13 19:01:42
|
|
do not align audio and video vpts on discontinuity, this finally makes multiangle
DVDs play smooth with my DXR3
CVS patchset: 3508
CVS date: 2002/12/13 16:47:41
|
|
CVS patchset: 3498
CVS date: 2002/12/12 22:54:57
|
|
CVS patchset: 3490
CVS date: 2002/12/12 01:58:37
|
|
segfault)
if out_samples were bigger than 2*in_samples.
CVS patchset: 3483
CVS date: 2002/12/11 16:22:05
|
|
=> you can install xine-lib 0.9.13, gnome-xine 0.1, xine-lib 1-beta*,
xine-ui 0.9.15 in that order and both applications will work
=> distributions can have libxine0 and libxine1 binary packages (not the
-dev/-devel) packages installed inparallel for a smooth upgrade path
I can't get rid of the feeling that I have broken something here...
CVS patchset: 3479
CVS date: 2002/12/11 01:04:01
|
|
i want to make sure we have a 1:1 mapping of sending and waiting for BUF_CONTROL_HEADERS_DONE.)
- bring asf demuxer into a saner state when seeking. that fixes the problem
of wrong audio pts and hanging audio_out on huge sleeps (at least for asf).
CVS patchset: 3465
CVS date: 2002/12/08 21:43:50
|
|
file size and modification date to ~/.xine/catalog.cache.
next time xine starts the first stage loader won't need to open any libraries,
reducing the loading time. the difference under gdb can be quite dramatic.
note1: xine still have to initialize all input and demuxers on startup.
note2: i have not added post plugins to cache yet.
CVS patchset: 3464
CVS date: 2002/12/08 20:53:02
|
|
CVS patchset: 3452
CVS date: 2002/12/06 21:37:18
|
|
- improves playback on slower systems since we avoid false triggering of still
frames and duplicating images
CVS patchset: 3447
CVS date: 2002/12/06 18:38:35
|
|
- these debug messages are not that interesting for the user
CVS patchset: 3441
CVS date: 2002/12/06 17:16:40
|
|
copy() is called if needed when frames are enqueued for displaying.
also frames are not "copied" if late (good for slow machines with xshm)
CVS patchset: 3435
CVS date: 2002/12/06 01:38:22
|
|
- add copy_called flag. this will help us to track usage of copy() method.
any driver which support copy() must set this flag (for each "copied" frame)
- increase interface version
CVS patchset: 3433
CVS date: 2002/12/06 01:30:49
|
|
CVS patchset: 3431
CVS date: 2002/12/06 01:13:15
|
|
a good thing)
CVS patchset: 3422
CVS date: 2002/12/04 04:11:01
|
|
CVS patchset: 3417
CVS date: 2002/12/03 21:59:46
|
|
CVS patchset: 3408
CVS date: 2002/12/01 17:04:30
|
|
CVS patchset: 3405
CVS date: 2002/12/01 15:11:47
|
|
CVS patchset: 3404
CVS date: 2002/12/01 15:11:02
|
|
- fix some bugs in plugin loader that prevented it from traversing subdirs
- implement some public API functions for post plugin handling here
CVS patchset: 3403
CVS date: 2002/12/01 15:10:04
|
|
CVS patchset: 3402
CVS date: 2002/12/01 15:07:35
|
|
CVS patchset: 3398
CVS date: 2002/12/01 14:52:55
|
|
CVS patchset: 3394
CVS date: 2002/12/01 01:32:47
|
|
audio drift the same way i did for video doesn't work, since sound card
won't play it slower or faster because of the vpts's.
with this patch i hope to have definitely separated the 2 sources of
audio drift: (1) nominal sample rate discrepancy to stream pts,
(2) sound card drift to system clock.
- revert feedback audio loop
- updated metronom's audio drift
- always do a full gap filling if needed
CVS patchset: 3392
CVS date: 2002/11/30 22:09:42
|
|
anyway as soon as a new stream starts or another decoder is needed
benefit: the dxr3 video decoder will not close the device after playing the logo and
therefore the image will stay on screen
CVS patchset: 3389
CVS date: 2002/11/29 17:25:26
|
|
CVS patchset: 3388
CVS date: 2002/11/28 18:49:10
|