Since matroskademux will attempt to push unaligned buffers,
downstream might have trouble with those, especially if downstream
uses ORC, such as audioconvert.
Ensure we push buffers aligned to the basic type at least for
those raw buffers.
https://bugzilla.gnome.org/show_bug.cgi?id=659798
... to at least having it trigger a/v synchronization, possibly without
using provided values which are still not considered sane
(as previously dropped).
... when operating in non slave mode, and reset if detected.
This should avoid some (large) bogus outgoing timestamp due to jumps
in rtp time, as result of PAUSE/PLAY or seek or ...
... at least if not syncing to NPT time:
* either sync using RTCP SR data (as currently)
* only perform the above once using initial RTCP SR packets
* discard RTCP and sync by equating provided stream's clock-base rtptime,
as provided by jitterbuffer (typically obtained from RTP-Info in RTSP).
Changed the ebml reader's gst_ebml_peek_id_length() function so
that it returns the actual reason for why the peek failed, instead
of (almost) always returning GST_FLOW_UNEXPECTED. This prevents
the pulling task from sending EOS when doing a flushing seek.
matroskademux performs segment tricks to skip gaps in streams,
notably at start for non 0 based files. There may however be
cases when full presentation (including intermediate gaps) is
desired, so a property allows to configure as of which gap
to act (or not at all).
API: GstMatroskaDemux::max-gap-time
Fixes#659009.
Subtract the first timestamp of a stream from all input buffers to
get 0-based timestamps for creating a sane ctts table. Without this
patch the ctts could have larger values than needed, causing the
playback to have a delay at startup.
As the first timestamp is only found after a few buffers are queued
(due to possible reordered buffers), once we find the first timestamp
we subtract it from all buffers on the queue, from that point on,
all buffers have their timestamps subtract when they are collected.
https://bugzilla.gnome.org/show_bug.cgi?id=658659
Frame duration might vary for 1 usecond, in this case matroskamux
decides to create BLOCKGROUP instead of SIMPLEBLOCK.
Convert duration to timecodescale which is (typically) less precise, and
then also allow the difference of 1/-1 to arrange for less sensitive check.
Based on patch by Alexey Fisher <bug-track@fisher-privat.net>
Fixes#653080.
Some encoders (Arecont) do not like the long OPTIONS sent at startup as sent by
GStreamer, but do accept the short header as sent by Live555.
This patch makes the extending the request optional by adding a property
(short-header).
Fixes#655805.
API: GstRTSPSrc:short-header
This likely breaks stuff. The good: all of the methods now create
field images aligned with input frames, without timestamp mangling.
The bad: this touches a lot of code, much of which is hairy and in
need of cleanup. However, at this point we can reasonably create a
PSNR-based test.
In particular, do so even if failing to read while prerolling,
such as when reading from a partial file (eg, while it is being
downloaded).
This fixes a wedge in playbin2.
https://bugzilla.gnome.org/show_bug.cgi?id=651965
Yes, I was tracking another bug and the small test file I generated
to test with improbably just happened to trigger this, with a second
and last frame of 1615 bytes.
https://bugzilla.gnome.org/show_bug.cgi?id=656649
We need to keep the lock held because we don't want a push before the "new-ssrc-pad"
handler has completed. But we may want to push an event from inside that handler, hence
the recursive mutex.
https://bugzilla.gnome.org/show_bug.cgi?id=650916
Some h264 payloaders are unfortunately buggy and don't correctly set the
E bit in FU-A NAL when they have ended. Work around this by assuming
such a fragmentation unit has ended when there was no packet loss and a
new NAL is started
When pushing out buffers over S/PDIF or HDMI, IEC 61937 payloading
requires each buffer to contain 6 blocks from each substream. This adds
code to collect all the frames needed to meet this requirement before
pushing out a buffer.
https://bugzilla.gnome.org/show_bug.cgi?id=650313
Using the current RTCP interval to timeout SSRC collision can lead to
collisions being timed out immediately if a BYE packet is sent because
it is sent immediately, so the interval is 0. This is not what we
want. So just set a static 10 times the default RTCP interval, it
should be enough
https://bugzilla.gnome.org/show_bug.cgi?id=648642
... which is particularly needed when merging NAL units, where not resetting
would lead to output of an older (pre-flush) AU (with unintended timestamp).
Current matroska demux calculates the pixel aspect ratio only if both
DisplayHeight and DisplayWidth are set, but it is legal to use only
one variable if the other is equal to PixelWidth or PixelHeight, at
least the mkclean utility is doing that. So this makse mkcleaned
files play correctly.
https://bugzilla.gnome.org/show_bug.cgi?id=654744
A missing sys/param.h include results in:
/usr/include/sys/proc.h:64: error: 'MAXLOGNAME' undeclared here (not in a
function)
/usr/include/sys/proc.h:285: error: 'MAXCOMLEN' undeclared here (not in a
function)
when compiling goom on openbsd/ppc. We can just remove the two sys/ includes
here, they are not needed for anything.
https://bugzilla.gnome.org/show_bug.cgi?id=654749
... introduced when shuffling around code for the async implementation
by setting state of source (and udp sources) in _play before downstream
flushing is undone.
The gst_base_parse_set_frame_rate call was predicated on a change to
sample rate, duration or profile. However, the block count per frame can
also change between packets, which would result in incorrect buffer
durations.
Some video frames, for example alt-ref frame in VP8, will be
never displayed. This is why it has duration=0.
This patch allow to use this duration.
Bug: 654175
Signed-off-by: Alexey Fisher <bug-track@fisher-privat.net>
Fixes: #652144.
gstudpnetutils.h:32:0: error: "WINVER" redefined
/usr/i686-w64-mingw32/sys-root/mingw/include/_mingw.h:231:0: note: this is the
location of the previous definition
Move the following function to matroska-read-common.[ch] from
matroska-demux.c and matroska-parse.c:
- gst_matroska_{demux,parse}_parse_attachments
https://bugzilla.gnome.org/show_bug.cgi?id=650877
Move the following function to matroska-read-common.[ch] from
matroska-demux.c and matroska-parse.c:
- gst_matroska_{demux,parse}_parse_attached_file
https://bugzilla.gnome.org/show_bug.cgi?id=650877
Move the following function to matroska-read-common.[ch] from
matroska-demux.c and matroska-parse.c:
- gst_matroska_{demux,parse}_parse_metadata_id_tag
https://bugzilla.gnome.org/show_bug.cgi?id=650877
Move the following function to matroska-read-common.[ch] from
matroska-demux.c and matroska-parse.c:
- gst_matroska_{demux,parse}_parse_metadata_id_simple_tag
https://bugzilla.gnome.org/show_bug.cgi?id=650877
AUTHOR only existed in an old version of the spec and ARTIST is
the new replacement for this. We are still reading both to still
be compatible with old files.
Fixes bug #644875.
Move the following functions to matroska-read-common.[ch] from
matroska-demux.c and matroska-parse.c:
- gst_matroska_{demux,parse}_get_seek_track
- gst_matroska_{demux,parse}_reset_streams
https://bugzilla.gnome.org/show_bug.cgi?id=650877
Move the following functions to matroska-read-common.[ch] from
matroska-demux.c and matroska-parse.c:
- gst_matroska_index_seek_find
- gst_matroska{demux,parse}_do_index_seek
https://bugzilla.gnome.org/show_bug.cgi?id=650877
Move the following function to matroska-read-common.[ch] from
matroska-demux.c and matroska-parse.c:
- gst_matroska_{demux,parse}_tracknumber_unique
https://bugzilla.gnome.org/show_bug.cgi?id=650877
It does not work at all (A/V sync issues), is not very useful,
other containers work much better with Dirac and Dirac in AVI
is not supported by other software.
Fixes bug #541215.
Move the following functions to matroska-read-common.[ch] from
matroska-demux.c and matroska-parse.c:
- gst_matroska_{demux,parse}_encoding_cmp
- gst_matroska_{demux,parse}_read_track_encodings
https://bugzilla.gnome.org/show_bug.cgi?id=650877
Move the following functions to matroska-read-common.[ch] from
matroska-demux.c and matroska-parse.c:
- gst_matroska_{demux,parse}_peek_id_length_pull
- gst_matroska_{demux,parse}_peek_id_length_push
https://bugzilla.gnome.org/show_bug.cgi?id=650877
Move the following functions to matroska-read-common.[ch] from
matroska-demux.c and matroska-parse.c:
- gst_matroska_{demux,parse}_encoding_order_unique
- gst_matroska_{demux,parse}_read_track_encoding
https://bugzilla.gnome.org/show_bug.cgi?id=650877
Move the following functions to matroska-read-common.[ch] from
matroska-demux.c and matroska-parse.c:
- gst_matroska_decode_content_encodings
- gst_matroska_decompress_data
https://bugzilla.gnome.org/show_bug.cgi?id=650877
Replace the following functions with their gst_matroska_read_common_*
counterparts:
- gst_matroska_{demux,parse}_parse_index
- gst_matroska_{demux,parse}_parse_skip
- gst_matroska_{demux,parse}_stream_from_num
Introduce GstMatroskaReadCommon to contain those members of
GstMatroskaDemux and GstMatroskaParse that were used by the above
functions.
https://bugzilla.gnome.org/show_bug.cgi?id=650877
Tell GstBaseParse the duration in samples instead of time, so that
a duration query in DEFAULT format will return the correct number
of samples without rounding errors. Baseparse will convert this
into time itself when needed.
https://bugzilla.gnome.org/show_bug.cgi?id=650785
When not using the fieldanalysis element immediately upstream of deinterlace,
behaviour should remain unchanged. fieldanalysis will set the caps and flags on
the buffers such that they can be interpreted and acted upon to produce
progressive output.
There are two main modes of operation:
- Passive pattern locking
Passive pattern locking is a non-blocking, low-latency mode of operation that
is suitable for close-to-live usage. Initially a telecine stream will be
output as variable framerate with naïve timestamp adjustment. With each
incoming buffer, an attempt is made to lock onto a pattern. When a lock is
obtained, the src pad and output buffer caps will reflect the pattern and
timestamps will be accurately interpolated between pattern repeats. This
means that initially and at pattern transitions there will be short periods
of inaccurate timestamping.
- Active pattern locking
Active pattern locking is a blocking, high-latency mode of operation that is
targeted at use-cases where timestamp accuracy is paramount. Buffers will be
queued until enough are present to make a lock. When locked, timestamps will
be accurately interpolated between pattern repeats. Orphan fields can be
dropped or deinterlaced. If no lock can be obtained, a single field might be
pushed through to be deinterlaced.
Locking can also be disabled or 'auto' chooses between passive and active
locking modes depending on whether upstream is live.
... since the TEARDOWN sequence might not have had a chance to even start,
but at least connections should be closed (synchronously) and state cleaned up.
See #632504.
Simplify the command handling; passing a command to thread means we really
want it to get the message, which means to always flush provided the command
can handle being interrupted. Command thread indicates whether command
allows interruption and ensure non-flushing connection as it subsequently
needs it.
In particular, this also makes the TEARDOWN sequence interruptable
and also prevents races where _loop_ could miss a command and would
continue receiving (or at least trying to).
See #632504.
With the async state changes, it is possible that we need to open the stream
before play and pause.
Also make sure we remember a previous open failure so that we don't keep trying
again.
Simplify the command handling, only continue looping when we have not received
another command or when the previous loop was successfull.
Avoid looping on a disconnected socket.
If the lock is not released before emitting a signal, it may cause a deadlock
if any other function in the element is called.
Also removed an unused timestamp parameter
https://bugzilla.gnome.org/show_bug.cgi?id=649617
Since the segment duration is given in terms of the
GST_MATROSKA_ID_TIMECODESCALE we should only convert it into
nanoseconds when we are sure that any scale specified in the file has
been read.
https://bugzilla.gnome.org/show_bug.cgi?id=650258
If the bitrates for all but one audio/video streams are known, and the
total stream size and duration can be determined, this calculates the
unkown bitrate as (stream size / duration) - (sum of known bitrates).
While this is not guaranteed to be very accurate, it should be good
enough for most purposes.
For example, this is useful for H.263 + AAC streams where no 'btrt' atom
is available for the video portion.
https://bugzilla.gnome.org/show_bug.cgi?id=619548
This parses the 'damr' atom if present, and exports the maximum bitrate
of the stream using the mode set field to determine the highest bitrate
frame type that might be present.
https://bugzilla.gnome.org/show_bug.cgi?id=620186
Since the segment duration is given in terms of the
GST_MATROSKA_ID_TIMECODESCALE we should only convert it into
nanoseconds when we are sure that any scale specified in the file has
been read.
https://bugzilla.gnome.org/show_bug.cgi?id=650258
Otherwise wavenc will fail if upstream decides to set equivalent caps or caps
with additional information later.
Thanks to Alexander Schremmer for finding this bug.
A duration tag gets inserted only for streamable=false, so only
update/write the duration later if we actually inserted that tag,
otherwise we write garbage into other tags.
https://bugzilla.gnome.org/show_bug.cgi?id=649060
Refuse h264 caps without stream-format and codec_data fields for
now, to avoid creating broken files. This might cause some pipelines
that worked previously to fail. However, the move from -bad to -good
is our only chance to fix this up, so make it strict for now. We can
always change it back to be less strict in future.
https://bugzilla.gnome.org/show_bug.cgi?id=647919
Don't use g_assert() for error handling, even if they're highly unlikely.
Either we *know* that something can't happen, in which case we
should just not handle it, or we think something can happen, but it is
very very unlikely that it will ever happen, in which case we should
handle it like any other error instead of asserting.
g_assert() is best left for conditions we have control of, like checking
internal consistency of our code, not checking return values of external
code.
Fixes a bunch of warnings when compiling with -DG_DISABLE_ASSERT:
gstrtpgsmpay.c: In function 'gst_rtp_gsm_pay_handle_buffer':
gstrtpgsmpay.c:130:17: warning: variable 'rtpgsmpay' set but not used
gstspeexenc.c: In function 'gst_speex_enc_encode':
gstspeexenc.c:904:19: warning: variable 'written' set but not used
pulsesink.c: In function 'gst_pulsesink_change_state':
pulsesink.c:2725:9: warning: variable 'res' set but not used
pulsesrc.c: In function 'gst_pulsesrc_change_state':
pulsesrc.c:1253:7: warning: variable 'e' set but not used
We use -DG_DISABLE_ASSERT for the pre-releases, which makes these
warnings pop up in cases that were previously covered by g_assert_not_reached()
and the like:
tvtime/greedyh.c:801:14: warning: 'scanline' may be used uninitialized in this function
matroska-mux.c:501:19: warning: 'context' may be used uninitialized in this function
GCC 4.6.x spits warnings about variables that are unused but set. Such
variables have been removed where trivial but with comments left behind
for informational purposes in some cases.
gst_rtp_session_chain_recv_rtcp () was changed in commit 490113d4
to always return GST_FLOW_OK instead of the return value of
rtp_session_process_rtcp (), so we'll keep it that way.
This is needed for automatic transcoding using encodebin. Our typefinder
does not always add a variant to the found caps, and encodebin needs
an *exact* match to the caps on the source pad template, so we need
to add the variant-less video/quicktime caps to the template as well
for encodebin to be able to find it. Add unit test for this as well.
https://bugzilla.gnome.org/show_bug.cgi?id=642879
... and not only when sort-of feeling like it.
In any case, if it turns out all really is in order,
and presumably DTS == PTS, then no ctts will be produced anyway.
That is, all sorts of problems arise with re-ordered input timestamps that
tend to defy automagic handling for every case, so allow for a few variations
that can be tried depending on circumstances.
Also try to document accordingly.
Also fixes#638288.