Re-work the dtsdec element code to unify it with changes made it a52dec,
including support for reverse playback and dynamic channel negotiation
on the source pad.
When updating a pad, send the update to half a second behind the SCR,
which avoids ever updating the start time for a pad to beyond the end of
the cell. Also, remember the last actual new-segment start time for each
pad, and use it when closing the segment.
When the current button number is higher than the number of available
buttons, switch to the highest numbered button rather than the lowest.
Also, don't throw errors when we fail to retrieve some button info
from libdvdnav, just reset the highlight.
Send the commands-changed navigation message when the set of available
DVD menu button actions changes, and handle the commands navigation
query so that (e.g.) Totem can know about the available navigation
commands.
Protect pad exposure with a preroll lock to avoid situations
where no-more-pads is fired more than once, or fired just before
the last pad is actually added.
Send segment updates to the audio and subpicture pads more frequently,
but less often to the video pad, where timestamps appear less often.
This helps with gap filling on some DVDs.
When creating a filler audio buffer in rsnaudiomunge, generate
a bit more, as audio sinks don't seem to preroll otherwise. This
needs a better algorithm in general, to intelligently fill the
gap, rather than hard-coding a value.
Make timed still frames work better by extending the current segment
when needed, and restarting the still sequence with the correct
remaining duration when the wait it interrupted by activation of a
highlight NAV packet.
The part number reported while in a menu doesn't reflect the selected
menu, so it's pointless to use it to report which menu we're in (Audio,
Angle etc). Just report "DVD Menu" in the title tag instead.
The logical audio and subpicture stream number doesn't always correspond
with the physical substream it is coming from. When configuring the demuxer
pads, use the mapping table provided in each PGC to get the layout and
ensure the demuxer creates the correct pads.
modplug's sndfile.h conflicts with libsndfile's sndfile.h, so
we'll access it directly using modplug/sndfile.h. Fixes#573849
Signed-off-by: David Schleef <ds@schleef.org>
No need to unref buffers if pad_push returns something non-FLOW_OK.
In fact, this will cause assertions or crashes. However, we do need
to unref the input buffer whenever we don't pass it downstream. Also,
a non-OK flow return is not an error, so don't post error messages on
the bus - the pipeline may just be shutting down. Miscellaneous other
clean-ups and crack removal. Plenty of work left for those who feel
like it.
