home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
ftp.cs.arizona.edu
/
ftp.cs.arizona.edu.tar
/
ftp.cs.arizona.edu
/
icon
/
newsgrp
/
group96a.txt
/
000192_icon-group-sender _Mon Aug 19 18:27:25 1996.msg
< prev
next >
Wrap
Internet Message Format
|
1996-09-05
|
1KB
Received: by cheltenham.cs.arizona.edu; Tue, 20 Aug 1996 08:08:03 MST
Date: Mon, 19 Aug 1996 18:27:25 -0500
From: jeffery@dragon.cs.utsa.edu (Clinton Jeffery)
Message-Id: <9608192327.AA12311@water.cs.utsa.edu>
To: cwfraser@microsoft.com
Cc: icon-group@cs.arizona.edu
In-Reply-To: <c=US%a=_%p=msft%l=RED-25-MSG-960819171536Z-30291@tide21.microsoft.com> (message from Chris Fraser on Mon, 19 Aug 1996 10:15:36 -0700)
Subject: Re: evaluating
Content-Length: 507
Errors-To: icon-group-errors@cs.arizona.edu
> During the evaluation of, say,
> every (1 to 1000000)/real(n) ...
> n is converted to floating point once, not one million times, right?
e1 / e2
are evaluated similar to
e1 & e2
which is to say, every time e1 produces another result e2 (which might be
a generator) is restarted.
A good optimizing compiler ought to be able to figure out whether n changes
in the do... clause of the every loop, and if not, skip re-evaluation, but
the semantics of the language (and icont...) say: re-evaluate.