I had little doubt my Marpa grammar had issues. I'm still trying to wrap my head around lexer vs. parser and how Marpa's two levels interact. I figured the main difference was captures vs. non capture in my reading. I didn't realize there was a performance impact. I'm still having a very difficult time understanding the differences between the G0/L0 rules.

I actually tried to use the :discard lexeme, but I found that I couldn't get the required newline after a partial line comment to match properly after this. They seemed to be discarded before they could match. Maybe I'll have another look.

Thanks for the input, it's greatly appreciated :) I'll look over your changes and play with the code to see what I can learn.