From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me
X-Spam-Level: 
X-Spam-Status: No, score=-1.3 required=5.0 tests=BAYES_00,INVALID_MSGID
	autolearn=no autolearn_force=no version=3.4.4
X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: 1094ba,9f0bf354542633fd
X-Google-Attributes: gid1094ba,public
X-Google-Thread: 103376,d901a50a5adfec3c
X-Google-Attributes: gid103376,public
From: William Clodius <wclodius@lanl.gov>
Subject: Re: Fortran or Ada?
Date: 1998/09/30
Message-ID: <36129598.2C67@lanl.gov>#1/1
X-Deja-AN: 396459912
Content-Transfer-Encoding: 7bit
References: <36068E73.F0398C54@meca.polymtl.ca>
 <slrn70dv6k.2qp.gisle@krake.ii.uib.no> <m3n27sqhms.fsf@visar.physast.uga.edu>
 <6u8r5o$aa4$1@nnrp1.dejanews.com> <m3zpbq24wq.fsf@visar.physast.uga.edu>
 <360A3446.8AD84137@lmco.com> <6udre0$ha1$1@nnrp1.dejanews.com>
 <19980925.185359.250@yktvmv.watson.ibm.com> <6uifdr$dog$1@nnrp1.dejanews.com>
 <19980928.184428.604@yktvmv.watson.ibm.com>
 <wclodius-2909981741110001@clodius.lanl.gov>
 <19980929.214309.386@yktvmv.watson.ibm.com> <6utg60$h6l$1@nnrp1.dejanews.com>
 <19980930.153012.753@yktvmv.watson.ibm.com>
Content-Type: text/plain; charset=us-ascii
Organization: Los Alamos National Lab
Mime-Version: 1.0
Newsgroups: comp.lang.fortran,comp.lang.ada
Date: 1998-09-30T00:00:00+00:00
List-Id: <comp.lang.ada>

jbs@yktvmv.watson.ibm.com wrote:
> <snip>
>          It appears you do not turn off the check, instead you "protect"
> the check which is expensive.
>  <snip>

No. You can turn off the check, but this implies that you are accepting
garbage as an output. Accepting garbage is rarely desirable, although it
might have been less harmful for the Ariane 5 than what actually
happened.

In order to preserve information, you protect the variables being
checked, which may in turn depend on the results of the check. In the
simplest case this would result in the variable being set to an extrema,
i.e., too large a value being set to the largest integer value, but this
too results in a loss of information. You know you overflowed but not by
how much. More sophisticated would be to replace the normal linear
function with a non linear function so that precision is lost as you
approach and surpass the overflow value for the original linear
transformation.

-- 

William B. Clodius		Phone: (505)-665-9370
Los Alamos Nat. Lab., NIS-2     FAX: (505)-667-3815
PO Box 1663, MS-C323    	Group office: (505)-667-5776
Los Alamos, NM 87545            Email: wclodius@lanl.gov