Hi Edward,
1. Is there a more elegant way of doing number parsing? In particular, are there token parsers that don't consume trailing whitespace, or is there a better way to do this with the primitives.
Parsec defines a combinator it calls 'lexeme' which the tokenizer wraps each of its functions in. The purpose of the tokenizer is to create a set of parsing combinators that ignore whitespace, comments, and some other handy stuff like checking for collisions with reserved keywords. To consume the trailing whitespace is not a bug, it's an abstraction layer, and Parsec is consistent about only using this abstraction in the Token module. It's too bad that the 'nat' function in Token is not defined in Parsec's Char module, and because of that, you need to copy-paste that code or roll your own.
It seems that the "token" approach of parsing lends itself to a different style of parsing than the one I'm doing
That's correct. Sounds to me like you shouldn't bother creating a tokenizer. You might even be able to get away with using the regex library instead of Parsec. -Greg