It improves EXT-X-TARGETDURATION when some chunk duration is bigger than requested -hls_time N
In the current implementation, it is ceiling the value (if float part is bigger than 0.001) but it should be rounded. Specification https://tools.ietf.org/html/draft-pantos-http-live-streaming-22#page-49 allows both implementation, but rounding is more precise and better matches user needs (especially having in mind that HLS segmenter could generate slightly longer than requested). See conversation https://trac.ffmpeg.org/ticket/6533 --- libavformat/hlsenc.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/libavformat/hlsenc.c b/libavformat/hlsenc.c index 74a3249b73..1c36ae4a12 100644 --- a/libavformat/hlsenc.c +++ b/libavformat/hlsenc.c @@ -1038,8 +1038,8 @@ static int hls_window(AVFormatContext *s, int last) goto fail; for (en = hls->segments; en; en = en->next) { - if (target_duration <= en->duration) - target_duration = get_int_from_double(en->duration); + if (target_duration < en->duration) + target_duration = (int)round(en->duration); } hls->discontinuity_set = 0; -- 2.11.0 (Apple Git-81) _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
