theora dev - Feb 2009 - Pages with granulepos = -1 while encoding..

Hi all !

Following my recent issued with my encoder, I've narrowed the issue to the 
fact that, when encoding, for some returned pages, the granulepos returned by 
the libogg is -1. 

Is that normal ? How should I intepret it if I want to order the pages ?


Romain

Large packets may span pages. -1 means no packet ends on that page.
It should be held till the next page with a >= 0 granpos comes, in which
case this granpos also logically applies to the preceding -1 pages.

On Thu, Feb 5, 2009 at 7:20 AM, Romain Beauxis <toots at rastageeks.org>
wrote:
> Following my recent issued with my encoder, I've narrowed the issue to
the
> fact that, when encoding, for some returned pages, the granulepos returned
by
> the libogg is -1.
Technically speaking, I think you can also use a greedy algorithm and
always write out granulepos=-1 pages as soon as you get them, since by
definition all you know is that they must come between two marked
pages from that same logical stream. The delay-until-the next
timestamp suggestion ogg.k made is more likely to be optimal in the
future though, if we ever block the packed+spanning packets case.

 -r