Fri Mar 23 20:47:42 CET 2012

Constraint programming, layout & choreography

The trick is to be able to non-causally describe convergence.  It's
easier to do this "backwards in time".  Same goes for spacial

Example: have two animiation meet in one point.  It's simpler to
directly specify that they meet at time t=t' than to figure out when
to start them such that this property is met implicitly.

Example: how to center an asymmetric animation based on its final

I need a way to convert a constraint-based description to a linear
sequence.  This most likely requires a 2-step approach: 1) solve all
unknowns and 2) perform a coordinate transformation or lookup.

So how to do this?  I'm probably going to be helped with just a local
constraint propagation solver (equations as bi-directional functions).