Return-Path: Received: from SEGATE.SUNET.SE by xiron.pc.helsinki.fi with smtp (Linux Smail3.1.28.1 #1) id m0t3tEJ-0000ZOC; Sat, 14 Oct 95 01:09 EET Message-Id: Received: from listmail.sunet.se by SEGATE.SUNET.SE (LSMTP for OpenVMS v1.0a) with SMTP id CC4B0C2F ; Sat, 14 Oct 1995 0:09:31 +0100 Date: Fri, 13 Oct 1995 14:57:19 -0400 Reply-To: Logical Language Group Sender: Lojban list From: Logical Language Group Subject: subscripts on numbers and lerfu X-To: lojban@cuvmb.cc.columbia.edu To: Veijo Vilva Content-Length: 1080 Lines: 22 >> 3) How does one say "any two of the man, the woman and the kid"? That >> is, how does one extract n elements from the ce-specified set? The >> best I could think of is "re lo cmima be le nanmu kuce le ninmu kuce >> le verba". Is there a way of doing it without cmima? Something with >> LU'A? I tried "re lu'a ny. ce ny. xire ce vy.", but parser won't >> allow that. > >The problem there appears to be with {xi}. The parser doesn't like it >after a BY. {re lu'a ny. ce ny.boi xire ce vy.} seems to work. And if one looks at non-terminal rules 300, 400, 812, and 817 in the YACC grammar, one will see that both numbers and lerfu strings must have a BOI between them and any subscript. This BOI is not normally elidable because the parser inserts the elided BOI BEFORE looking at the subscript XI, and an elliptical BOI doesn't include a subscript. Why are the rules written this way? The syntax of multiple subscripts is SO messy that it was easy to get ambiguous grammars. Could it be fixed? Perhaps, but it is not considered an important problem. lojbab