[antlr-interest] wildcard in tree grammar

[Gavin Lambert]

A few minutes ago, I wrote:
 >certain possible input trees simply won't be able to
 >be parsed any more (or at least not as nicely).

Well, ok, that's a bit too much of an exaggeration; after all, 
this:

   ^('not' ^('and' c51=. c52=.))

with "." handling both cases (as I still maintain it should) could 
be rendered like this if there were separate . and _ operators 
(which I still dislike):

   ^('not' ^('and' c51=wild c52=wild))

wild : . | _ ;

But that seems really silly.