On 18/05/14 14:58, Martin Storsjö wrote:
> On Sun, 18 May 2014, Luca Barbato wrote:
>
>> The marker #EXT-X-DISCONTINUITY had been introduced in the specification
>> to support intermixing different sources (e.g. advertisement).
>> ---
>> libavformat/hls.c | 2 ++
>> 1 file changed, 2 insertions(+)
>>
>> diff --git a/libavformat/hls.c b/libavformat/hls.c
>> index 290f12e..827abbb 100644
>> --- a/libavformat/hls.c
>> +++ b/libavformat/hls.c
>> @@ -278,6 +278,8 @@ static int parse_playlist(HLSContext *c, const
>> char *url,
>> } else if (av_strstart(line, "#EXTINF:", &ptr)) {
>> is_segment = 1;
>> duration = atof(ptr) * AV_TIME_BASE;
>> + } else if (av_strstart(line, "#EXT-X-DISCONTINUITY", &ptr)) {
>> + c->first_packet = 1;
>> } else if (av_strstart(line, "#", NULL)) {
>> continue;
>> } else if (line[0]) {
>> --
>> 1.9.0
>
> Well, sure we can parse the #EXT-X_DISCONTINUITY marker, but what actual
> effect does this have on receiving those streams? Setting
> c->first_packet doesn't really do anything (it only makes sure we
> recheck the discard flags immediately).
I'm not sure if I had been lucky or not, but hls w/out this patch gets
royally confused when you scramble segments, with it it behaves as expected.
> What one might want to do, though, is to make sure we don't compare
> timestamps from one variant from before the discontinuity to timestamps
> from another one - however I think that's not much of an issue right now
> either since we enforce returning packets according to segment sequence
> number before we check the timestamps themselves.
I thought setting first packet would take care of it (since surprisingly
the testcase I created worked fine that way), I'll try to figure out if
it was just a lucky situation for some reasons.
lu
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
