Comment by ajross

You're not discussing in good faith. The performance comparison was to the "parse to the end" variant that you suggested as equivalent. The natural way you implement that (again, very simple) algorithm wants the early exit, yes, for obvious performance reasons.

We're done. You're "not convinced" my way is simpler because you're not willing to give ground at all. This is a dumb thing to argue about. Just look at some historical parsers for similar languages, I guess. Nothing I've said here is controversial at all.