My main concern is that while you'll get error messages from parse failures, they may not actually do a good job of pointing to the cause of the failure. I don't know much about parsing, but I gather that a lot of people have put a lot of work into designing parsing frameworks with decent error reporting, and you're not likely to get anything near that by grafting errors onto a more primitive system.
Well, I once tried implementing a parser that evaluates integer
addition, multiplication, exponential, tetration, pentation, and so on
infinitely. The operators were + with precedence 6, * with precedence
7, ^ with precedence 8, ^^ with precedence 9 (for tetration), ^^^ with
precedence 10 (for pentation), ^^^^ with precedence 11 (for hexation),
and so on infinitely. I've not succeeded implementing it using
ordinary ReadP.
Libraries mailing list
Libraries@haskell.org
http://mail.haskell.org/cgi-bin/mailman/listinfo/libraries