Thanks, I was mildly disappointed that I couldn't use that regexp
feature in lex-abbrevs.
I did end up using the arbitrary code feature in the "lexer" actions to
great effect: to scan a string literal token that requires paren
balancing, and do to the unescaping in the same pass. The "lexer"
syntax nicely gives you the input port, with the port position is
exactly where you'd expect it to be.
I was also briefly disappointed that the parser syntax didn't support
some EBNF conveniences. But as soon as I started adding actions to
build my AST, I found that I didn't want EBNF anyway. The result was
some elegant use of simple list-building and functional hashes.
Overall, a positive experience with the "parser-tools" collection.
--
http://www.neilvandyke.org/
____________________
Racket Users list:
http://lists.racket-lang.org/users
