That link is a very high quality write up.

I’m having a hard time understanding how your “haystack compilation” would work. I get the desire to be able to efficiently run many regexes on the same large file, but how could you optimize the file for arbitrary regex matching?