[ogg-dev] Seeking to granules in discontinuous streams

[ogg.k.ogg.k at googlemail.com]

> do you still think you need all this, if you are allowed to have equal
> granulepos on subsequent pages?

The time figures I mentioned earlier were without bits reserved for the
counter, so yes, I think so.

However, I'm not sure about allowing the same granulepos for distinct
events at the same time for the following reason:

The Seek Algorithm bisects to the latest page with a granpos equal or
lower than the one we're seeking to. If you've got, say, two events at
granule 10, and the bisection finds the second one, it'll then bisect to
the later side, so it'll seek to that second page, missing the first. That
seems like the wrong thing to do. In my case, I wouldn't want that to
happen.

Something else that needs more thinking about.