If this is so, http://pastebin.com/WjDP1xmj is not representative for your computer language. Since there is only a table, the tokenizer takes apart the lexical elements and since most of the lexical elements are fixed, gperf is perfect for that job. I don’t see at what point you need to deal with many alternatives.
Fill up the various possible actions with some code && you’ll see where your FSM got stuck , and/or generate graphviz output and follow the states by hand for a given input string. In case line 23 of your sample input data should be (possibly partially) matched by your “raiseaction” I’d start debugging by looking at what the “any” machine really matches jg
<<wlEmoticon-smile[1].png>>
_______________________________________________ ragel-users mailing list [email protected] http://www.complang.org/mailman/listinfo/ragel-users
