Newsgroups: rec.arts.int-fiction
Path: nntp.gmd.de!news.rwth-aachen.de!news-koe1.dfn.de!main.Germany.EU.net!Germany.EU.net!EU.net!howland.erols.net!news.bbnplanet.com!cam-news-hub1.bbnplanet.com!uunet!in2.uu.net!192.35.48.11!hearst.acc.Virginia.EDU!murdoch!not-for-mail
From: nr@adder.cs.Virginia.EDU (Norman Ramsey)
Subject: Re: Specific Inform criticism (Long)
X-Nntp-Posting-Host: adder-fo.cs.virginia.edu
Message-ID: <5an0kp$gvg@adder.cs.Virginia.EDU>
Sender: usenet@murdoch.acc.Virginia.EDU
Organization: University of Virginia
References: <59g6p8$grj@bartlet.df.lth.se> <ant282255313M+4%@gnelson.demon.co.uk> <5a748r$iir@life.ai.mit.edu> <ant3021181cbM+4%@gnelson.demon.co.uk>
Date: Sun, 5 Jan 1997 01:37:29 GMT
Lines: 25

In article <ant3021181cbM+4%@gnelson.demon.co.uk>,
Graham Nelson  <graham@gnelson.demon.co.uk> wrote:
>[A spirited and interesting defense of Inform, including...]

 >>      So the language has two separate notions of "property" (the existence
 >>      of which is already a source of pointless non-orthogonality), one
 >>      with an ad hoc magic-number limit.
 >
 >No, I really am not going to beat my breast over this one.  It's an
 >implementation matter, really.  You just don't need to use "Property":
 >it's hardly mentioned in the manuals any more.  I think the parallel
 >is with C, where some variables can be flagged as "register" to ask
 >the compiler to compile code using them more efficiently.  Of course,
 >there is necessarily a magic-number limit to the number of registers
 >in the machine.  In Inform 6, "Property" is a directive used by the
 >library to improve the efficiency of code, that's all.

May I humbly suggest that a future version of the Inform compiler do
what good C compilers have long done, i.e., analyze the code to decide
which properties are best given this special representation?  Even if
`Property' must be left in for backwards compatibility, it can be a
no-op. 
-- 
Norman Ramsey
http://www.cs.virginia.edu/~nr
