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=-0.3 required=5.0 tests=BAYES_00,
	REPLYTO_WITHOUT_TO_CC autolearn=no autolearn_force=no version=3.4.4
X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: 103376,32c6d2fb891775fc
X-Google-Attributes: gid103376,public
X-Google-ArrivalTime: 2003-10-28 06:14:54 PST
Path: 
 archiver1.google.com!news2.google.com!news.maxwell.syr.edu!newsfeed.stueberl.de!newsfeed.freenet.de!news.tu-darmstadt.de!fu-berlin.de!uni-berlin.de!adsl-213-200-246-247.cybernet.CH!not-for-mail
From: Vinzent 'Gadget' Hoefler <ada.rocks@jlfencey.com>
Newsgroups: comp.lang.ada
Subject: Re: Modular integers
Date: Tue, 28 Oct 2003 15:13:25 +0100
Organization: JeLlyFish software
Message-ID: <bnltkl$11i07f$1@ID-175126.news.uni-berlin.de>
References: <bnlmg1$kqv$04$1@news.t-online.com>
 <Pftnb.86846$5n.57615@bignews5.bellsouth.net>
 <bnloh2$q3d$03$1@news.t-online.com>
Reply-To: v.hoefler@acm.org
NNTP-Posting-Host: adsl-213-200-246-247.cybernet.ch (213.200.246.247)
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: quoted-printable
X-Trace: news.uni-berlin.de 1067350485 35193071 213.200.246.247 (16 [175126])
X-Newsreader: Forte Agent 1.8/32.548
Xref: archiver1.google.com comp.lang.ada:1774
Date: 2003-10-28T15:13:25+01:00
List-Id: <comp.lang.ada>

Lars wrote:

>Why couldn't the compiler insert code that checks what the upper limit =
of
>the type is,

It does, if necessary. AFAICS in your case it isn't: The upper limit
of your subtype is the same as that of the base type which is a
modular type.

So where would you expect the overflow? You don't get rid of the
modulo-operation just by defining a sub type.

>then substracts the current value of the variable from the
>upper limit to see if the difference is bigger or equal to the number =
you
>want to add? If it is not, it could throw an exception ...

Why should it? No thinkable result of the modulo 2**64 operation would
overflow your new range constraint for the subtype:

IOW: "(A + 1) mod 2**64 in 0 .. 2**64 - 1" is always true.

But try this with a subtype with a range only up to 2**64 - 2, this
*will* give you the expected overflow:

|with Ada.Text_IO;
|
|procedure t is
|   type    X_Base is mod 2**64;
|   subtype X_1    is X_Base range 0 .. 2**64 - 1;
|   subtype X_2    is X_Base range 0 .. 2**64 - 2;
|
|   A : X_Base;
|   B : X_1;
|   C : X_2;
|
|begin
|   A :=3D X_Base'Last;
|   Ada.Text_IO.Put (X_Base'Image (A));
|   A :=3D A + 1;
|   Ada.Text_IO.Put (X_Base'Image (A));
|  =20
|   B :=3D X_1'Last;
|   Ada.Text_IO.Put (X_Base'Image (B));
|   B :=3D B + 1;
|   Ada.Text_IO.Put (X_Base'Image (B));
|  =20
|   C :=3D X_2'Last;
|   Ada.Text_IO.Put (X_Base'Image (C));
|   C :=3D C + 1;
|   Ada.Text_IO.Put (X_Base'Image (C));
|end t;

compiled with GNAT3.15p gives:

| 18446744073709551615 0 18446744073709551615 0 18446744073709551614
|
|raised CONSTRAINT_ERROR : t.adb:25 range check failed

Isn't that the expected result?


Vinzent.