It sounds like you have an illegible, compact regex in mind. If you write regexes like that for others to read, I feel sorry for the other people. If you write

It sounds like you have an illegible, compact regex in mind. If you write regexes like that for others to read, I feel sorry for the other people. If you write the regex like you would production code, i.e. with whitespace, newlines, and comments[1], I imagine 200 lines of regex would be about the same as 200 lines of any other type of production code, be it APL, Haskell, or conjunctive normal form. I realise this is a bit of a circular argument: if you write the code such that 200 lines is about n units of review in a specific language, then 200 lines will be about n units of review in any language. But that's also kind of the point. There's a very specific logic density that you can accomplish for production code to be readable, and that results in about that density being reviewed. [1]: In Python you can do this with the re.VERBOSE flag, in other languages you can do it by breaking the regex apart into separate variables that are then combined together.