This makes both the lexer and parser quite a bit easier to use. Counting column numbers isn't also really needed when parsing XML/HTML.