Return-Path: <jimc@math.ucla.edu>
	(Sendmail 5.61/1.07) id AA05308; Mon, 12 Apr 93 08:11:20 -0700
Message-Id: <9304121511.AA05308@julia.math.ucla.edu>
Subject: Re: Lujvo paper, part 3.1 
             <9304121415.AA03662@julia.math.ucla.edu> 
Date: Mon, 12 Apr 93 08:11:20 -0700
From: jimc@math.ucla.edu
X-Mozilla-Status: 0011

A little behind-the-scenes sniping:

> There will be cases where the component gismu have many places in common,
> and each gismu orders these places differently. To be consistent with GDS,
> we choose the rightmost gismu in the equivalent tanru as the gismu whose
> place structure is used. This is because it is closest to being the
> head of the tanru, the tertanru, on which the place structure of a tanru
> is based.

Such a situation is a sign that the gismu places may need to be
reordered. The places of the Lojabn gismu have been gone over pretty
thoroughly, with a lot of attention to "optimizing" the use of each
gismu in isolation, but very little thought to assembling them in lujvo
-- the dikyjvo orientation.  I warned Lojbab that he was painting
himself into a corner by proceeding forward in gismu places without
making the necessary foundation of dikyjvo rules.  He did go forward,
and now there's a lot of text that would be invalidated by the
necessary place changes.

Anyway, this section of the paper is good.  

		-- jimc