It seems that filter lexers overwrite the default nextToken() method, and in so doing, no check is made for skipped tokens, rendering the skip() command useless -- is this a bug, or a feature?