Implement keyframe seeking in oggdemux by doing the double seek trick. First
seek to the required position, then read pages for all streams to grab the
granulepos (to know the timing of the keyframe) of each stream, then seek back
to the first keyframe.
After a seek, discard all packets before the packet with the granulepos on it so
that the output buffers contain valid timestamps.
Reorder some code so that we check the timestamps before allocating and pushing
an output buffer.
Do more checks on valid packets in ogm mode.
Do some additional checks on the granulpos timestamp before using it for
calculating the duration because oggdemux generates wrong granulepos now.
Fixes seeking somewhat again.
OGM demuxing no longer requires helper elements. It's done internally
in oggdemux. Vorbis comments are still not handled because I don't
have anything to test with.
gstoggstream.c:419: error: format ‘%lld’ expects type ‘long long int’, but argument 8 has type ‘gint64’
gstoggdemux.c:2253: error: format ‘%lld’ expects type ‘long long int’, but argument 8 has type ‘GstClockTime’
gstoggdemux.c:2333: error: format ‘%lld’ expects type ‘long long int’, but argument 8 has type ‘GstClockTime’
This reverts commit 60aa09d28c.
First drawing the real text and then the outline produces ugly
text in lower resolutions. The outline line width needs to be somehow
changed relative to the resolution. Fixes bug #602924.
Timestamp generation was broken by the last commit for formats
with a non-zero granule shift. Also keep track of the last keyframe
so that we can regenerate granulepos for theora.
Add a granule to granulepos conversion function. Fix the duration
function for vorbis. Handle timestamps on header packets differently
and be more careful about calculating OFFSET and OFFSET_END. After
this change, timestamps for vorbis don't exactly match up with the
timestamps that vorbisparse outputs, but it's unclear if vorbisparse
is actually correct and it would add a lot more code to make oggdemux
match vorbisparse. Fixes#602790.
Adds code that parses headers of various formats encapsulated in
Ogg in order to calculate timestamps and durations of each buffer.
Removes the creation of helper decoder elements to do this calculation
via conversion queries.
Fixes: #344013, #568014.
There's not much point in using GST_DEBUG_FUNCPTR with GObject
virtual functions such as get_property, set_propery, finalize and
dispose, since they'll never be used by anyone anyway. Saves a
few bytes and possibly a sixteenth of a polar bear.
Previously, the code always rounded to even sizes. Now it only ensures
that pic_x and pic_y are multiples of 2 if the output format requires
it.
Also inlcudes fixes to take pic_x/y into account properly when copying
the buffer.
https://bugzilla.gnome.org/show_bug.cgi?id=594729
warning: format '%llu' expects type 'long long unsigned int', but argument 8 has type 'GnomeVFSFileOffset'
warning: format '%lld' expects type 'long long int', but argument 9 has type 'guint64'
This guarantees that people who use theoraenc without modifying any
properties will end up with a reasonably good quality output.
48 is also the default of the encoder_example application shipped with
libtheora.
Revert previous 'fix' for bug #588717 and fix it properly, whilst
maintaining the streamheader field on the output caps. Also make
sure we don't leak header buffers we couldn't push when downstream
is unlinked. Add unit test for the presence of the streamheader
field on the output caps and for the issue from bug #588717.
Fix caps warning when there's no element linked downstream, and pass
not-linked flow return value correctly up the chain, so we error out
correctly. Fixes#588717.
The GInputStreams are now requested by a vfunc from
the subclasses instead of relying that the subclass
sets it until it's needed.
This might also fix bug #587896.
POSIX and your local friendly ctime(3) manual entry says that localtime_r isn't
required to set the state variables that define the current timezone. Indeed,
glibc (at least 2.9) doesn't do this for subsequent calls. The effect is that
if the system timezone is changed for a running program between two calls to
gst_clock_overlay_render_time, it won't be noticed. For glibc, changing the
timezone equals /etc/localtime being modified.
Fixes bug #587676.
Try to guess from the paranoia-mode setting whether playback or
ripping is wanted, and use a smaller cache size if we're likely
to be doing playback, to avoid a long startup delay. Since this
was the value used in older cdparanoia versions, it should be
fine in any case. See #586331.
This setting was added in cdparanoia 10.2. The default value is good
for audio extraction, but lower values (previous versions of cdparanoia
used 150) are better for realtime playback.
Fixes#586331.
For this to work properly, theoradec and vorbisdec need to put
tag events received from upstream into the pending_events list
so they get pushed out after any newsegment event, not before.
Bitrates are stored as 32-bit signed integers in the vorbis
identification headers, but seem to be read incorrectly,
namely as unsigned 32-bit integers, into the vorbis structure
members which are of type long, which makes our check for
values <= 0 fail with files that put -1 in there for unset
values.
Previous history is in Schroedinger. Depends on, and is an example
of using, GstBaseVideo* base classes.
Code was reindented, and an #ifdef HAVE_ENCODER removed.
When we feed the ogg sync layer, we need to feed it contiguous data even if the
sync layer did not consume all of it yet. This makes sure that it always finds
the next page even for more corrupted files. Use a different read_offset for
this purpose. since we now keep track of the sync layer, we don't have to reset
after finding a start of a page.
Add some more debug info for the error paths.
Only reset the sync layer when we perform a seek operation.
Avoid failure when the next chain has no bos pages but instead simply ignore it.
when we receive unknown page serial numbers mid stream, don't fail but post a
warning and hope that we get back on track later.
Fixes#579642
Add the start position of the first segment to the running time
used to generate buffer timestamps in vorbisenc. This avoids generating
buffers which fall outside the initial segment. The element segment
handling requires more extensive fixing, but this at least prevents
regressions. Fixes: #580020
Add property "speed-level" to control the amount of motion searching
the encoder does. This is only available in libtheora >= 1.0 and
will silently fail with earlier libraries. Fixes: #572275.
Signed-off-by: David Schleef <ds@schleef.org>
This is due to race conditions between functions that
modified the mixer like set_volume and
snd_mixer_handle_events since the handle_events
can now be called at any time.
Fixed by adding locking around any snd_mixer call
since even read functions can modify the mixer stucture, since
alsa likes to clear it's values before reading new ones.
The favorite race condition seemed to be that set_volume
called read_elem (in alsalib) that reset the volumes to
0 and then read them with read_x_volume. This read looped
on each channel and as the race condition occured the
channels value could be anything , most of the time
it was 0. Thus no value was read or only the value of
one channel was and the volume was reset to 0.
Fixes bug #478512.
Original commit message from CVS:
* configure.ac:
* ext/gio/gstgio.c: (plugin_init):
Add plugin dependency for the GIO and GVfs modules.
Fixes bug #566876.
Original commit message from CVS:
* configure.ac:
* ext/gnomevfs/gstgnomevfs.c: (plugin_init):
Add plugin dependency for the gnomevfs modules.
Fixes bug #566875.
Original commit message from CVS:
* configure.ac:
* ext/libvisual/visual.c: (plugin_init):
Use new core API to make registry re-scan the plugin
whenever visualisations are added or removed (see #350477).
Original commit message from CVS:
* ext/vorbis/vorbisenc.c: (gst_vorbis_enc_base_init),
(gst_vorbis_enc_init):
Make vorbisenc's pad template behave like vorbisdec's. Fixes a leak with
pad templates.
Original commit message from CVS:
* ext/ogg/gstoggdemux.c: (gst_ogg_pad_class_init),
(gst_ogg_pad_dispose), (gst_ogg_pad_finalize):
Use G_DEFINE_TYPE for the OggPad to get some threadsafe type
init from glib.
Original commit message from CVS:
* ext/ogg/gstoggdemux.c: (gst_ogg_demux_sink_event):
If no stream was found before receiving EOS, post an error message.
Fixes#561924.
Original commit message from CVS:
* ext/theora/gsttheoraenc.h:
* ext/theora/theoraenc.c: (gst_theora_enc_init),
(theora_buffer_from_packet), (theora_push_packet),
(theora_enc_sink_event), (theora_enc_is_discontinuous),
(theora_enc_chain):
Parse segment events.
Pass incomming buffer timestamps to outgoing buffers.
Use the running_time to construct the granulepos.
Fixes#562163.
Original commit message from CVS:
* ext/gio/gstgiobasesrc.c: (gst_gio_base_src_create):
* ext/gnomevfs/gstgnomevfssrc.c: (gst_gnome_vfs_src_create):
Use gst_buffer_try_new_and_alloc() and fail properly if the
allocation failed. This prevents abort() if downstream elements
request an insane amount of memory.
Original commit message from CVS:
* ext/ogg/gstoggdemux.c: (gst_ogg_pad_submit_packet),
(gst_ogg_demux_deactivate_current_chain),
(gst_ogg_demux_perform_seek), (gst_ogg_demux_handle_page),
(gst_ogg_demux_loop):
* ext/ogg/gstoggdemux.h:
Copy seqnums around to track playback segments and messages.
Original commit message from CVS:
Based on patch by: Matthias Kretz <kretz at kde dot org>
* ext/alsa/gstalsasink.c: (gst_alsasink_open),
(gst_alsasink_prepare), (gst_alsasink_unprepare),
(gst_alsasink_write):
Make all access non-blocking so that we can better handle unplugging
of usb devices. Fixes#559111
Original commit message from CVS:
Based on patch by: Robin Stocker <robin at nibor dot org>
* ext/theora/gsttheoradec.h:
* ext/theora/theoradec.c: (gst_theora_dec_init),
(theora_dec_setcaps), (theora_handle_type_packet),
(theora_dec_decode_buffer), (theora_dec_change_state):
Parse input caps and make the PAR override the encoded PAR when
specified by a container. Fixes#555699.
