| Message ID | 20201004142841.179499-1-derek.buitenhuis@gmail.com |
|---|---|
| State | Accepted |
| Commit | 2be3eb7f77a3b6433016f5691ed504f99e0377da |
| Headers | show |
| Series | [FFmpeg-devel] Allow using only the mfra info for seeking using the fragment index | expand |
| Context | Check | Description |
|---|---|---|
| andriy/default | pending | |
| andriy/make | success | Make finished |
| andriy/make_fate | success | Make fate finished |
On 04/10/2020 15:28, Derek Buitenhuis wrote: > From: Justin Ruggles <justin.ruggles@gmail.com> > > The mfra has enough information to enable seeking, and reading it is > behind an AVOption flag, so we shouldn't require that sidx information > also be present in order to seek using the fragment index. > > Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com> > --- > This is especially important want I/O is networked, since otherwise it will > read essentially the entire file by seeking to every sidx in the file, which > for a live-style FMP4 can be a *lot*. > > Dale Curtis is CC'd, as this is probably highly relevant to him. > > Note: If a global sidx is present, that information will override the mfra, but > if sidx are before each moof, those will no longer be parsed in mov_read_header > if you set use_mfra_for, which can result in a wrong duration since mfra isn't > used for setting duration while sidx is. > --- > libavformat/mov.c | 1 + > 1 file changed, 1 insertion(+) Ping. - Derek
On 07/10/2020 16:44, Derek Buitenhuis wrote: >> The mfra has enough information to enable seeking, and reading it is >> behind an AVOption flag, so we shouldn't require that sidx information >> also be present in order to seek using the fragment index. >> >> Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com> >> --- >> This is especially important want I/O is networked, since otherwise it will >> read essentially the entire file by seeking to every sidx in the file, which >> for a live-style FMP4 can be a *lot*. >> >> Dale Curtis is CC'd, as this is probably highly relevant to him. >> >> Note: If a global sidx is present, that information will override the mfra, but >> if sidx are before each moof, those will no longer be parsed in mov_read_header >> if you set use_mfra_for, which can result in a wrong duration since mfra isn't >> used for setting duration while sidx is. >> --- >> libavformat/mov.c | 1 + >> 1 file changed, 1 insertion(+) > > Ping. Will push late today, or tomorrow, if nobody objects. - Derek
On 08/10/2020 11:26, Derek Buitenhuis wrote: >> Ping. > > Will push late today, or tomorrow, if nobody objects. Pushed. - Derek
diff --git a/libavformat/mov.c b/libavformat/mov.c index 4f64e96bc0..7fd43a8fc5 100644 --- a/libavformat/mov.c +++ b/libavformat/mov.c @@ -7529,6 +7529,7 @@ static int mov_read_mfra(MOVContext *c, AVIOContext *f) goto fail; } while (!ret); ret = 0; + c->frag_index.complete = 1; fail: seek_ret = avio_seek(f, original_pos, SEEK_SET); if (seek_ret < 0) {