From cbmvax!uunet!hplb.hpl.hp.com!kers Thu Jul 11 09:02:22 1991 Return-Path: (15.11/15.6+ISC) id AA08794; Thu, 11 Jul 91 12:11:21 bst Date: Thu, 11 Jul 91 12:11:21 bst From: Chris Dollin Message-Id: <9107111111.AA08794@tigger.hpl.hp.com> To: lojban-list@snark.thyrsus.com Subject: the grammar Status: RO The recently-posted grammar had some constructs in it (such as ``&'', ``...'') for which I am unclear as to precedence. (I proved that ``&'' was associate, which helps.) I am *assuming* that ``...'' takes the *shortest* construct to the left as its operand and that ``&'' binds more tightly than ``|''; also that ``#'' is effectively just a non-terminal name (ie it could have been a rule called ``possible-frees''). Since ``[foo ...]'' seems to mean ``zero or more foo's'', could one have defined a suffix operator ``*'' for this? (Then ``#'' becomes ``free*''.) If I have time, I may stuff this grammar through my I-don't-care-two-hoots- whether-it's-ambiguous-I'll-just-give-you-all-the-parses parser. (Unfortunately it doesn't handle null productions, so a little re-working would be needed.) Regards, | It's strange how you suddenly discover bizarre and dangerous Kers. | holes in your knowledge. I fell down one today. - Steve Knight