Original commit message from CVS:
* gst/qtdemux/qtdemux.c: (qtdemux_parse_trak):
Fix for obvious mistake, where we first shift the offset and then
read a samplesize element assuming the old offset. Note that this
part still has something weird, i.e. my movies containing those
don't actually play well, but at least there's something that looks
like sound now.
Original commit message from CVS:
* gst/asfdemux/gstasfdemux.c: (gst_asf_demux_video_caps),
(gst_asf_demux_setup_pad):
Use 25fps as our "fake" fps value (marked for fixage in 0.9.x)
instead of 0. Reason is simple: some elements have a fps range
of 1-max instead of 0-max. So now ASF video actually works.
Original commit message from CVS:
* gst/sine/gstsinesrc.c: (gst_sinesrc_get):
return taglist correctly from _get function, don't gst_pad_push it.
(fixes#137042)
Original commit message from CVS:
* ext/alsa/gstalsamixer.c: (gst_alsa_mixer_free_list):
* ext/alsa/gstalsamixertrack.c: (gst_alsa_mixer_track_class_init),
(gst_alsa_mixer_track_new):
* ext/alsa/gstalsamixertrack.h:
Fix ancient leftovers... MixerTrack is a GObject.
Original commit message from CVS:
* ext/alsa/gstalsa.c: (gst_alsa_class_probe_devices):
* sys/oss/gstosselement.c: (gst_osselement_class_probe_devices):
Don't block during probing...
Original commit message from CVS:
* ext/alsa/gstalsa.c: (gst_alsa_get_type), (gst_alsa_class_init),
(gst_alsa_get_property), (gst_alsa_probe_get_properties),
(gst_alsa_class_probe_devices), (gst_alsa_class_list_devices),
(gst_alsa_probe_probe_property), (gst_alsa_probe_needs_probe),
(gst_alsa_probe_get_values), (gst_alsa_probe_interface_init),
(gst_alsa_open_audio), (gst_alsa_close_audio):
* ext/alsa/gstalsa.h:
Add propertyprobe interface implementation, add some device-name
property, all this so that it looks good in gnome-volume-control.
Original commit message from CVS:
* configure.ac: the Hermes library controls hermescolorspace, not
colorspace.
* ext/mpeg2dec/gstmpeg2dec.c: (gst_mpeg2dec_base_init),
(gst_mpeg2dec_init): minor pet peeve: disable code with #ifdef,
not /* */
* ext/sdl/sdlvideosink.c: Change XID to unsigned long.
* ext/sdl/sdlvideosink.h: ditto.
* gst/colorspace/gstcolorspace.c: Fix old comments about Hermes
Original commit message from CVS:
* gst-libs/gst/xoverlay/xoverlay.c: (gst_x_overlay_set_xwindow_id),
(gst_x_overlay_got_xwindow_id):
* gst-libs/gst/xoverlay/xoverlay.h:
replace XID with unsigned long to get rid of the xlibs dependency in
XOverlay (fixes#137004)
Original commit message from CVS:
Port all elements that can be ported to videofilter, and fix up the caps.
Can someone with a big-endian machine please check them?
Original commit message from CVS:
* sys/oss/gstosssink.c: (gst_osssink_init), (gst_osssink_get_time),
(gst_osssink_chain), (gst_osssink_change_state):
Latest fixes for A/V sync, audio playback and such. This is about
all... MPEG playback issues are mostly related to the async build-
up of MPEG files, I cannot fix that. Use basicgthread to solve it.
Original commit message from CVS:
* ext/divx/gstdivxdec.c:
Downgrade priority. We prefer ffdec_mpeg4.
* ext/faad/gstfaad.c: (gst_faad_srcgetcaps), (gst_faad_srcconnect),
(gst_faad_chain), (gst_faad_change_state):
Fix capsnego. Doesn't work for some sounds because we don't have
a 5:1 to stereo element.
* ext/xvid/gstxvid.c: (plugin_init):
Add priority.
* sys/oss/gstosssink.c: (gst_osssink_init), (gst_osssink_chain),
(gst_osssink_change_state):
Add discont handling.
Original commit message from CVS:
* ext/mikmod/gstmikmod.c: (gst_mikmod_init), (gst_mikmod_loop),
(gst_mikmod_change_state):
* ext/mikmod/gstmikmod.h:
make mikmod's loop function not loop infinitely and call
gst_element_yield anymore
* gst/modplug/gstmodplug.cc:
fix pad negotiation
Original commit message from CVS:
* ext/lcs/Makefile.am: Fix so that the lcs colorspace plugin
doesn't conflict with the internal colorspace plugin.
* gst-libs/gst/audio/make_filter: Use `` instead of $() to
satisfy the crappy-ass shell shipped by a certain vendor.
* gst/videofilter/make_filter: same (bug #135299)
Original commit message from CVS:
* gst/qtdemux/qtdemux.c: (gst_qtdemux_add_stream),
(qtdemux_parse_trak):
Fix crash (j might be greater than n_samples, in which case we're
writing outside the allocated space for the array) and memleak.
Original commit message from CVS:
* sys/oss/gstosssink.c: (gst_osssink_chain):
And another caller that couldn't handle delay < 0 (unsigned
integer overflow). Video now continues playing on an audio
buffer underrun, and the clock continues working. Audio still
stalls.
Original commit message from CVS:
* sys/oss/gstosssink.c: (gst_osssink_get_delay),
(gst_osssink_get_time):
get_delay() may return values lower than 0. In those cases, we
should not actually cast to *unsigned* int64, that will break
stuff horribly. In my case, it screwed up A/V sync in movies
in totem rather badly.
