buffer during entropy decoding.

The pointer address could overflow, which would likely segfault. Instead set
the context error flag to indicate that the decoder tried to read past the
end of the packet data.

avoids implementation-defined unsigned-to-signed conversion and simplifies
the bounds checking.

The variable is not used anymore at that point.

This prevents errors and/or invalid writes in the next decode call due to
s->samples still being negative.

break some excessively long lines and remove space after '*'

This is already checked by avcodec_decode_audio3().

Specifically av_update_cur_dts(), av_seek_frame_binary() and
av_gen_search().

They are not supposed to be called outside lavf.

The documentation for CODEC_CAP_PARSE_ONLY and AVCodecContext.parse_only
indicates that they are utilized through avcodec_parse_frame(), which was
never actually implemented.

Its functionality was removed several years ago, so it doesn't do anything.
AVCodecContext.frame_number could serve the same purpose if someone
wants to debug the frame count.

On frame decoding failure, return an error if the frame is the same size as
the whole packet, otherwise just log an error message and return the number
of bytes consumed.

Specify avoptions and how to use pactl to list devices.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>

It currently use the simple api and is using the latency information
provided only to offset the stream start.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>

this was forgotten when the encoder was removed

avcodec_decode_audio3() already checks it before sending the packet to the
decoder.

Use AVCodecContext.frame_number instead.

This is already checked by avcodec_decode_audio3()

prior to decoding.

blocks.

The values will always be the same, so this change eliminates an unneeded
variable. It also gets rid of the need to reset src when memcpy() is used.