From richard@rrbcurnow.freeuk.com Wed Aug 29 15:01:57 2001 Return-Path: X-Sender: richard@rrbcurnow.freeuk.com X-Apparently-To: lojban@yahoogroups.com Received: (EGP: mail-7_3_2); 29 Aug 2001 22:01:57 -0000 Received: (qmail 59922 invoked from network); 29 Aug 2001 21:51:57 -0000 Received: from unknown (10.1.10.142) by l7.egroups.com with QMQP; 29 Aug 2001 21:51:57 -0000 Received: from unknown (HELO scrabble.freeuk.net) (212.126.144.6) by mta3 with SMTP; 29 Aug 2001 21:51:57 -0000 Received: from du-008-0092.freeuk.com ([212.126.151.92] helo=rrbcurnow.freeuk.com) by scrabble.freeuk.net with esmtp (Exim 3.12 #2) id 15cDFY-0002ra-00 for lojban@yahoogroups.com; Wed, 29 Aug 2001 22:51:53 +0100 Received: from richard by rrbcurnow.freeuk.com with local (Exim 2.02 #2) id 15cCDX-0006IS-00; Wed, 29 Aug 2001 21:45:43 +0100 Date: Wed, 29 Aug 2001 21:45:43 +0100 To: lojban@yahoogroups.com Subject: Re: [lojban] LALR1 question Message-ID: <20010829214543.A9524@rrbcurnow.freeuk.com> Mail-Followup-To: lojban@yahoogroups.com References: <20010828024050.A941@twcny.rr.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2i-nntp In-Reply-To: <20010828024050.A941@twcny.rr.com>; from rob@twcny.rr.com on Tue, Aug 28, 2001 at 02:40:50AM -0400 From: Richard Curnow I don't think the joi-overloading between selbri and sumti is the worst problem by any means. The most problematic case I came across when doing the pre-parser in jbofi'e was the constructions like I JA optional-simple-tense BO I JA optional-simple-tense KE I JA optional-simple-tense something-else where the decision to reduce or not depends on whether bo, ke or something else comes at the end of a tense which has potentially arbitrary length. So in this case no value of k is high enough for LR(k) to be up to the job. On Tue, Aug 28, 2001 at 02:40:50AM -0400, Rob Speer wrote: > On Mon, Aug 27, 2001 at 05:29:11PM -0600, Jay Kominek wrote: > > My conclusion: If you want the language to be syntactically unambiguous, > > LALR(1) is a fairly good choice. The most you'd want to do is switch to an > > LR(2) parser. If you need more than that, you're doing something wrong. > > Actually, now I'm wondering - would changing the language to LR(2) actually > help? What if you change {le broda joi le brode} to {le broda ui joi le brode} > - would that not parse in LR(2), or does UI somehow not count in the > lookahead? Am I looking at this all wrong? -- R.P.Curnow,Weston-super-Mare,UK | C++: n., An octopus made by http://www.rrbcurnow.freeuk.com/ | nailing extra legs on a cat.