I was unable to complete any meaningful part of Day 17 in Dust, and ended up with a purely JavaScript solution, which can be found here.

I was expecting this to be a challenge. One of the examples that motivated Dusa’s initial design was Adam Smith’s Map Generation Speedrun with Answer Set Programming. That example uses a half-assed inefficient version of Dijkstra’s algorithm that sort of horrified me when I came across it.

I’ve got ideas about how Dusa could be made to handle efficient minimization searches like for Dijkstra’s algorithm in the future. But for now, the language is not capable of handling that kind of search for a Day 17 AoC-sized problem.