Hacker News Reader

I’ll be the guy who postscripts that “hardmode” regexes would have important mathematical and computer science implications when/if they solve certain problems (like identifying primes). So it’s not pure pedantry, and there’s an important difference between “regexes to solve programming problems” and “regexes solving math problems”.

View on HN

"hardmode" regexes do have important computer science implications: they can be matched orders of magnitude faster [0].

Demonstrating the power of a regex engine by solving a problem like this is interesting and very clever, but it’s hopefully not used in production.

[0] https://swtch.com/~rsc/regexp/regexp1.html

Yeah I think we agree here. I didn’t mean to assert that chomskian regular grammars or regexes satisfiable by a state machine hadn’t already been useful; I meant to say that because they have existing math implications and properties that new uses for them would imply connections to other maths, analogous computations, new proofs or ways to solve things, etc.

And yeah, they’re faster because they can be statically compiled to the point of running strings “through” the expression rather than running the expression “on” the string, so to speak. (i.e as an FSM, iterating char by char the operations boil down to purely compare and branch to known destinations)

Reply on news.ycombinator.com