Looks good.
I’ll try and get to the bottom of the slow performance and problem with Grab on the Mac.
Yes, this is the expected behaviour. When you adjust sliders over time, these adjustments and their order are not stored by Grasshopper, only their current value.
For many types of optimisation this doesn’t affect the end result - if you have a catenary mesh and you increase the load and then the stiffness, or the other way around, or apply the full value of both at once, Kangaroo will still converge to the same solution.
However, with something like this pathfinding question, there are many nearby local energy minima, so which one it converges to will be affected by the way the strengths are varied over time.
I would like to do something to enable repeatability of this sort of timing-dependent optimisation.
I was sharing some thoughts about this recently here: Triggering a button - #8 by DanielPiker
With the new graph mapper functionality @DavidRutten is adding for GH2 it looks like making this sort of timeline control might get a lot easier.