silmeth 🇺🇦 on Nostr: npub129gva…e4kwg Doing linear scan of bytes instead of split(',') brings part1 ...

npub129gvast08lj986yftn7q5qlnj8yfqufxx0m33s9u5xssjm8c64rsve4kwg (npub129g…4kwg) Doing linear scan of bytes instead of split(',') brings part1 benchmark down to 33 µs. So yeah, I’m definitely losing some time there in parsing.

Actually, I noticed my parsing times going slightly up since I started using lines() and split() more liberally (I did hand-roll simple top-bottom parses for the first few puzzles and it was generally faster).