[MPlayer-users] Multi-file .ogm XCD problem

[Michael Behrisch]

On Mon, Jan 19, 2004 at 02:13:56PM +0000, Martin Collins wrote:
> On Mon, 19 Jan 2004 14:20:04 +0100
> Michael Behrisch <behrisch at informatik.hu-berlin.de> wrote:
> 
> > The code in scan_stream is more or less the one which was
> > build_syncpoint_table before. Could someone please check 
> > whether this error also occurs in the older version with -forceidx
> > enabled because then it is not my fault :-)
> 
> 1.0pre2 works without -forceidx. With -forceidx I get
> 
> Building syncpoint table 100%
> Bad page sync while building syncpoints table (2176)
> Bad page sync while building syncpoints table (4096)
> [many repeats]
> Bad page sync while building syncpoints table (4096)
> Bad page sync while building syncpoints table (260)
> Building syncpoint table 195%
> Bad page sync while building syncpoints table (1894)
> Bad page sync while building syncpoints table (4096)
> [many repeats]
> Building syncpoint table 204%
> 
> after which I give up as it has taken so long. It does look like the
> same problem as with 1.0pre3 and no -forceidx, just with messages.
> I assume -forceidx sets index_mode = 2 in which case this makes perfect
> sense.
> 
> I tried enclosing the whole while(1) loop in if(index_mode==2) and it
> works. Dunno if that is a fix or just a bodge. Seems to make more sense
> than always building the syncpoint table but only displaying the messages
> when -forceidx is used. It does mean syncpoint table building is broken
> though.

No this does not fix it, it simply disables my patch. The syncpoint table
is not always build, I just reused the scanning code for my own purposes,
but I would very much like the author of the original syncpoint table code
to comment on this. Maybe there is something wrong in checking for the end 
of the stream. There was a similar report (OGG on XCD) in February 2003,
http://article.gmane.org/gmane.comp.video.mplayer.user/12638/
dont know if this is related.

Michael