[MPlayer-dev-eng] [PATCH] Fix compilation against latest FFmpeg.

[Alexander Strasser]

On 2021-05-01 13:04 +0200, Alexander Strasser wrote:
[...]
> Revision after the Major Bumps
> ------------------------------
>
>     commit 8b83a4a885e3b92bd09354d6cb182fca47f80973
>     CommitDate: Fri Apr 30 10:24:32 2021 +0200
>
>         avutil/mem: Also poison new av_realloc-allocated blocks
>
> Now I get more than 500 checksum changes on `make fatetest`, lots
> of them look worrying because there are simply no decoded frames
> and at all.
>
> This is probably related with failures like this:
>
>     ./mplayer "$FATE_SAMPLES"/h264-conformance/MR8_BT_B.h264
>
>     MPlayer UNKNOWN-9 (C) 2000-2021 MPlayer Team
>
>     Playing /home/beastd/devel/fate/samples//h264-conformance/MR8_BT_B.h264.
>     libavformat version 59.0.100 (internal)
>     libavformat file format detected.
>     [Eval @ 0x55e772033360]Undefined constant or missing '(' in 'keepside'
>     [h264 @ 0x55e771fa0920]Unable to parse option value "keepside"
>     [h264 @ 0x55e771fa0920]Error setting option fflags to value +keepside.
>     LAVF_header: av_open_input_stream() failed
>
>
>     Exiting... (End of file)
>
> Others are worrying because previously differently decoded frames become
> all the same, like in this example:

Unsure at what cost, but simply removing this

    av_dict_set(&opts, "fflags", "+keepside", 0);

line in libmpdemux/demux_lavf.c at L608 makes the tests happy again.


[...]

  Alexander