[FFmpeg-cvslog] r20839 - trunk/libavformat/id3v2.c

[Reimar Döffinger]

On Mon, Dec 14, 2009 at 11:17:07AM -0800, Baptiste Coudurier wrote:
> On 12/14/2009 11:13 AM, Reimar D?ffinger wrote:
> >On Mon, Dec 14, 2009 at 12:23:39PM +0100, Michael Niedermayer wrote:
> >>diff -urN /home/michael/ffmpeg-revs/ffmpeg-r20539/trunk/libavformat/gxf.c ./gxf.c
> >>--- /home/michael/ffmpeg-revs/ffmpeg-r20539/trunk/libavformat/gxf.c	2009-12-13 14:13:21.000000000 +0100
> >>+++ ./gxf.c	2009-11-15 04:50:20.000000000 +0100
> >>@@ -464,7 +464,7 @@
> >>          if (skip)
> >>              url_fskip(pb, skip);
> >>          pkt->stream_index = stream_index;
> >>-        pkt->dts = field_nr;
> >>+        pkt->dts = field_nr*st->time_base.num;
> >>          return ret;
> >>      }
> >>      return AVERROR(EIO);
> >
> >Huh? Seems like complete nonsense to me, if the time base was e.g.
> >3/25 then the minimum delay between two consecutive frames would become
> >9/25 seconds?!?
> 
> Huh ?
> If timebase is 1001/60000, and field_nr is 2, 4, 6, 8.

That would then make dts
2002, 4004, ...
or in seconds
2*1001*1001/60000 = 33.4
4*1001*1001/60000 = 66.8
or in other words over 30 seconds per frame.
What am I missing??