Hi Jon,
Thanks for the answer. Then I have to leave this approach for a while.
Is there anyone working on the current autoplacer or is there something
planned? My intention was to test how differential evolution optimizer
from SciPy would work for this.
The usage case would be making "bundles" of footprint for preliminary
grouping before manual layout. I use the same for routing e.g. do a
preliminary layout and look for the bottlenecks from thin track
autorouter results, then do the final manually. Works pretty well for
dense discrete smd boads, where the main goal is to do everything in one
layer without vias.
-Vesa