[FFmpeg-devel] [PATCH 3/4] avformat: make flush_packets a tri-state and set it to -1 (auto) by default

[Michael Niedermayer]

On Mon, Jun 19, 2017 at 12:02:53AM +0200, Marton Balint wrote:
> If flushing is not disabled, then mux.c will signal the end of the packets with
> an AVIO_DATA_MARKER_FLUSH_POINT, and aviobuf will be able to decide to flush or
> not based on the preferred minimum packet size set by the used protocol.
> 
> Signed-off-by: Marton Balint <cus at passwd.hu>
> ---
>  doc/formats.texi            |  7 ++++---
>  libavformat/mux.c           | 12 +++++++++---
>  libavformat/options_table.h |  4 ++--
>  3 files changed, 15 insertions(+), 8 deletions(-)
> 
> diff --git a/doc/formats.texi b/doc/formats.texi
> index c51d4086db..ddd7743548 100644
> --- a/doc/formats.texi
> +++ b/doc/formats.texi
> @@ -182,9 +182,10 @@ Default is 0.
>  Correct single timestamp overflows if set to 1. Default is 1.
>  
>  @item flush_packets @var{integer} (@emph{output})
> -Flush the underlying I/O stream after each packet. Default 1 enables it, and
> -has the effect of reducing the latency; 0 disables it and may slightly
> -increase performance in some cases.
> +Flush the underlying I/O stream after each packet. Default is -1 (auto), which
> +means that the underlying protocol will decide, 1 enables it, and has the
> +effect of reducing the latency, 0 disables it and may increase IO throughput in
> +some cases.
>  
>  @item output_ts_offset @var{offset} (@emph{output})
>  Set the output time offset.
> diff --git a/libavformat/mux.c b/libavformat/mux.c
> index e1e49a81be..7c97b77881 100644
> --- a/libavformat/mux.c
> +++ b/libavformat/mux.c
> @@ -479,8 +479,10 @@ static int write_header_internal(AVFormatContext *s)
>          s->internal->write_header_ret = ret;
>          if (ret < 0)
>              return ret;
> -        if (s->flush_packets && s->pb && s->pb->error >= 0 && s->flags & AVFMT_FLAG_FLUSH_PACKETS)
> +        if ((s->flush_packets == 1 || s->flags & AVFMT_FLAG_FLUSH_PACKETS) && s->pb && s->pb->error >= 0)
>              avio_flush(s->pb);
> +        if (s->flush_packets && !(s->oformat->flags & AVFMT_NOFILE) && s->pb && s->pb->error >= 0)
> +            avio_write_marker(s->pb, AV_NOPTS_VALUE, AVIO_DATA_MARKER_FLUSH_POINT);


[...]
>      if (s->pb && ret >= 0) {
> -        if (s->flush_packets && s->flags & AVFMT_FLAG_FLUSH_PACKETS)
> +        if (s->flush_packets == 1 || s->flags & AVFMT_FLAG_FLUSH_PACKETS)
>              avio_flush(s->pb);
> +        if (s->flush_packets && !(s->oformat->flags & AVFMT_NOFILE))
> +            avio_write_marker(s->pb, AV_NOPTS_VALUE, AVIO_DATA_MARKER_FLUSH_POINT);

are these inteded to be if() instead of else if() ?

Either way iam in favor of this patch

thx

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

If you fake or manipulate statistics in a paper in physics you will never
get a job again.
If you fake or manipulate statistics in a paper in medicin you will get
a job for life at the pharma industry.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 181 bytes
Desc: Digital signature
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20170619/d343e5a6/attachment.sig>