The first frame pts must be saved until we have 8 frames since RoQ audio
requires 8 frames in the first packet.

This new API stores useful data in a dedicated structure
and has clearly delimited init functions.

Hopefully, uses of the old API can be replaced quickly.

Author of the code from qatar is Ronald S. Bultje
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Allow to specify the display aspect ratio adopted for rendering
subtitles.

Previously it was using the same value of the input video DAR, which is
inconsistent with most implementations.

Fix trac ticket #1098.

Check on < 0 rather than on != 0, this is more correct as a positive
error code from av_parse_ratio() value doesn't mean an error.

Signed-off-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

This is how it is defined in Amiga Developer CD from year 1992 and
is consistent with files created with ADPro.

Signed-off-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

* qatar/master: (27 commits)
  avconv: free packet in write_frame() when discarding due to frame number limit
  FATE: use +/- flag option syntax for vp8 emu-edge tests
  lavf: make av_interleave_packet_per_dts() private.
  lavf: deprecate av_read_packet().
  oggdec: output correct timestamps for Vorbis
  avconv: pass input stream timestamps to audio encoders
  lavc: shrink encoded audio packet size after encoding.
  xa: set correct bit rate
  xa: do not set bit_rate, block_align, or bits_per_coded_sample
  xa: fix end-of-file handling
  xa: fix timestamp calculation
  bink: fix typo in FFALIGN() argument
  bink: align plane width to 8 when calculating bundle sizes
  doc: pass -Idoc texi2html and texi2pod
  doc: texi2pod: add -I flag
  movenc: Add a min_frag_duration option
  rtsp: Set the default delay to 0.1 s for the RTSP/SDP/RTP demuxers
  libavformat: Set the default for the max_delay option to -1
  Generate manpages for AV{Format,Codec}Context AVOptions.
  doc/avconv: remove entries for AVOptions.
  ...

Conflicts:
	doc/Makefile
	doc/ffmpeg.texi
	doc/muxers.texi
	ffmpeg.c
	libavcodec/Makefile
	libavcodec/options.c
	libavcodec/vp8.c
	libavformat/options.c
	tests/fate/demux.mak
	tests/ref/fate/truemotion1-15
	tests/ref/fate/truemotion1-24

Merged-by: Michael Niedermayer <michaelni@gmx.at>

Update FATE references due to encoder delay.

Encoder output is delayed by several frames, so we keep a queue of input
frame timing info to match up with corresponding output packets.

FATE reference updated due timestamp rounding because of resampling from
44100 Hz to 16000 Hz in avconv.

Update FATE references due to encoder delay.

Reported-by: Deamon404
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

This simplifies matching of timestamps between input frames and output
packets.

Fixes a memleak when using the -frames option with audio.

Fixes Ticket1109

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

There is no reason for it to be public, it's only meant to be used
internally.

The caller can achieve the same effect (i.e. getting raw unparsed/mangled
packets) with av_read_frame() and AVFMT_FLAG_NOPARSE |
AVFMT_FLAG_NOFILLIN

Takes encoder delay into account by comparing first the coded page
duration with the calculated page duration. Handles last packet duration
if needed, also by comparing coded duration with calculated duration.
Also does better handling of timestamp generation for packets in the
first page for streamed ogg files where the start time is not
necessarily zero.

5 FATE test references updated due to using demuxer-generated timestamps that
are either not sample-accurate or are slightly off in the input file.

Also fixes stream duration calculation.

The values in the header refer to decoded data, not compressed data.

Do not output an extra packet when out_size is reached.
Also return AVERROR_EOF instead of AVERROR(EIO).

The packet duration is always 28 samples.