[MPlayer-cvslog] CVS: main/DOCS/tech mpcf.txt,1.67,1.68

[Michael Niedermayer]

Hi

On Thursday 03 March 2005 20:42, D Richard Felker III wrote:
> On Thu, Mar 03, 2005 at 06:16:53PM +0100, Michael Niedermayer CVS wrote:
> > CVS change done by Michael Niedermayer CVS
> >
> > Update of /cvsroot/mplayer/main/DOCS/tech
> > In directory mail:/var2/tmp/cvs-serv23240
> >
> > Modified Files:
> >  mpcf.txt
> > Log Message:
> > partial indexes
> > comments (with alternative suggestions and advantages) welcome
>
> wow! nice design!
> of course i'd expect no less from michael..

:)

i expected to be flamed ...


>
> > +index_distance
> > + distance at which indexes are approximately stored, or 0 if there are
> > + no indexes in the file
> > + in every [x*index_distance, (x+1)*index_distance) interval, there
> > + must be an index packet for every stream, and these packets must be
> > + located prior to all frames within the interval
> > + reasoning: this ensures good error recovery as there are many and
> > + evenly distributed indexes, and also allows very quick finding of the
> > + index packets
> > +
> > +index_modulo
> > +index_id
> > + each index packet contains every index_moduloth's keyframe of a stream,
> > + so the i'th keyframe of a stream will be at least in packets with
> > + (i % index_modulo) == index_id
> > +
> > +index_increment
> > + the index_id increment value, this MUST be a relative prime to
> > + index_modulo
>
> I'm confused what index_increment is used for.

hmm, not much, theres probably no need to store it in the file at all


>
> > + index_increment / index_modulo SHOULD be approximately 2/(sqrt(5)+1)
>
> And confused why this should be true..

so that the entries of n consecutive index packets are evenly distributed, 
note though i dont have a proof that the golden ratio is the optimal value 
here, i just think it is ...

[...]

-- 
Michael

"nothing is evil in the beginning. Even Sauron was not so." -- Elrond