i ran some comparisons on state representation width - 16-bit state IDs fit noticeably better into CPU cache than wider ones, and if you’re hitting 64K+ states you’re probably better off splitting into two simpler patterns anyway. one design decision i’m happy with is that when the engine hits a limit - state capacity, lookahead context distance - it returns an error instead of silently falling back to a slower algorithm. as the benchmarks above show, “falling back” can mean a 1000x+ slowdown, and i’d rather you know about it than discover it in production. RE# will either give you fast matching or tell you it can’t.
Tc′(x)x=Tc(x/s)x/s\frac{T_c'(x)}{x} = \frac{T_c(x / s)}{x / s}xTc′(x)=x/sTc(x/s)。新收录的资料对此有专业解读
。新收录的资料是该领域的重要参考
blog feed rm https://news.ycombinator.com/rss，这一点在新收录的资料中也有详细论述
01:47, 7 марта 2026Мир