DEV Community

Discussion on: 100 Languages Speedrun: Episode 65: Randomized Finite Automaton for Fast Thue Interpreter in Crystal

Collapse
 
taw profile image
Tomasz Wegrzanowski

Oh for sure. I think the biggest performance savings would come from switching from processing characters to processing bytes, as in UTF-8 this works perfectly well without any changes, and then we could just use 256 entry arrays instead of hashes for the trie.

Not like this really makes much difference for programs with so few rules.