PS:

Deliberately not considered (or rather, not included) are aspects of exact
positioning and playback. Addressing the former would have meant to litter
the document with more or less meaningless offset attributes having made-up
values. The latter topic has already been opened up into exciting new
directions in Joe's original proposal, which is why I chose to not further
weigh down the already lengthy encoding and to keep any ideas that I might
have in that area for a later time.