From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.5-pre1 (2020-06-20) on
	ip-172-31-74-118.ec2.internal
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 required=3.0 tests=BAYES_00 autolearn=ham
	autolearn_force=no version=3.4.5-pre1
Date: 4 Jan 93 23:19:46 GMT
From: noc.near.net!inmet!spock!stt@uunet.uu.net  (Tucker Taft)
Subject: Re: Ada Readability, Book Reference
Message-ID: <1993Jan4.231946.28620@inmet.camb.inmet.com>
List-Id: <comp.lang.ada>

In article <1993Jan4.174424.5699@seas.gwu.edu> 
  mfeldman@seas.gwu.edu (Michael Feldman) writes:

>Hmmm. How on earth could 1.0/3.0 be exact? What would the value be?
>
>I checked my Annotated LRM (thanks to Karl Nyberg!) and found AI-00209:
>"An implementation can refuse to evaluate a static universal real expression
>only if there are insufficient resources to evaluate the expression exactly,
>e.g., if there is insufficient memory available. Inexact results must not
>be delivered."
>
>This is interesting. I take it, then, that 1.0/3.0 will be flagged by
>the compiler, because - unless a rational representation is being
>used, which is always possible, I suppose - there aren't enough bits
>in a megabyte to evaluate it exactly. Do compilers use rationals
>for this?  I'll be back later, after a test or two of this!

All Ada compilers I know of have "infinite precision" rational
arithmetic packages for doing static universal real expressions.
So yes, 1.0/3.0 is computed exactly in the front end of the
compiler when it is static universal.  

Most Ada compilers compute *all* static real 
expressions (even non-universal) to "infinite precision" since
that is allowed by the accuracy model and is often simpler 
than simulating the target floating-point hardware in a cross-compiler.

The infinite precision rational arithmetic package is one of 
several things that makes an Ada front end a "fun" piece
of code to write and maintain ;-).

>Mike Feldman

S. Tucker Taft   stt@inmet.com
Intermetrics, Inc.
Cambridge,  MA  02138