Otherwise we will happily index into the array at NULL if the requested
index is smaller than the number of languages that were previously
allocated.
https://bugzilla.gnome.org/show_bug.cgi?id=777937
An asf object can't realistically be bigger than 2**32 bytes.
If it reports a size bigger than that, consider it corrupt and properly
propagate the error back.
https://bugzilla.gnome.org/show_bug.cgi?id=775070
In some corrupted files, we could end up with no actual streams
being exposed.
In those cases, make sure we properly propagate the failure all
the way to the loop function. This avoids ending up in cases where
we are neither EOS'd nor ERROR'd out from a pipeline point of view.
https://bugzilla.gnome.org/show_bug.cgi?id=774846
Fix seeking when demuxer is in INDEX state. This happens when we
reached the end of the stream. It should still be possible to do
a flushing seek and seek back to any other position though. Instead
the demuxer would just go straight to EOS again instead of going
back to processing packets again from the new position.
https://bugzilla.gnome.org/show_bug.cgi?id=749066
Support reverse playback for ASF format.
Parse packets and queue the payloads, then push
the payload buffer to decoder in reverse order.
Video buffers are pushed from KeyFrame to next
Keyframe.
https://bugzilla.gnome.org/show_bug.cgi?id=757341
We don't want to unconditionally send an EOS event downstream, but might want
to send a SEGMENT_DONE message/event instead if a segment seek was done
before.
https://bugzilla.gnome.org/show_bug.cgi?id=755469
When discovering a particular asf file, caps structure doesn't have
codec_data, and this was not being checked before using the same,
resulting in assertion error
https://bugzilla.gnome.org/show_bug.cgi?id=752803
Add parsing of stereoscopic metadata, and place into the caps to
the decoder.
Fix parsing of Advanced Mutual Exclustion objects.
https://bugzilla.gnome.org/show_bug.cgi?id=711190
Based on a patch by HyeJin Choi <meeshel78@hotmail.com>
We should give a chance to upstream to handle TIME seek events before trying
to handle it in asfdemux.
Fix a bug where fast backward seeks where discarded right away because they
were not supported by asfdemux.
https://bugzilla.gnome.org/show_bug.cgi?id=734471
For bytestream we don't want to expose it as codec_data but rather as
streamheader as it is not out-of-band data but data that should be
prepended to the beginning of the stream before the other buffers.
https://bugzilla.gnome.org/show_bug.cgi?id=735070
Relying on the gst_segment_do_seek results for getting the
seeking time should not be used if seek type is SET because the
do_seek function makes the 'cur` position of the seek normalized
on the segment range. This means that seeking beyond the end of the
file might not result in instant EOS as it will end up seeking to
the final ts of the file.
In asf seeking to the final timestamp can lead to seeking to the last
keyframe, meaning a few seconds of video will still be played, which should
not be the desired behavior when seeking beyong file limits.
