but then I'd have to parenthesise all my test names
-
this isn't *strictly* true (I mean you can parse any language with a composition of those functions, even if they're pure, for given definitions of those functions). Like we *could* build a ruby parser where the lexer was a pure scanner, but it would be very hard.
"can't" is probably too strong of a word to use here (clang doesn't use the lexer hack AFAIK), but most want to have typedefs be a separate token class
