If so, it'd be more interesting if the lexer didn't barf on syntax error, but perhaps just skipped characters until it finds a valid token. And probably some similar "best effort" algorithm for the parser.