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: 103376,bebb281f52290dd4
X-Google-Attributes: gid103376,public
From: Simon Wright <simon@pogner.demon.co.uk>
Subject: Re: Constraints?
Date: 1998/12/25
Message-ID: <x7vhfukit5q.fsf@pogner.demon.co.uk>#1/1
X-Deja-AN: 425684341
X-NNTP-Posting-Host: pogner.demon.co.uk:158.152.70.98
References: <x7v7lvkk7f8.fsf@pogner.demon.co.uk>
 <75ntvd$bue$1@tsunami.traveller.com>
X-Complaints-To: abuse@demon.net
X-Trace: news.demon.co.uk 914570909 nnrp-08:12025 NO-IDENT
 pogner.demon.co.uk:158.152.70.98
Organization: At Home
Newsgroups: comp.lang.ada
Date: 1998-12-25T00:00:00+00:00
List-Id: <comp.lang.ada>

"David C. Hoos, Sr." <david.c.hoos.sr@ada95.com> writes:

> Simon Wright wrote in message ...
> >package foo is
> >    type t is abstract tagged null record;
> >    type i (f : access t'class) is abstract tagged limited null record;
> >    type n is new t with null record;
> >    type r (f : access n) is new i (f) with null record;
> >    -- OA says foo.ads: Error: line 5 col 37 LRM:3.7(15), Subtype of
> >    -- discriminant must be statically compatible with subtype of the
> >    -- corresponding parent discriminant
> >end foo;
> The real problem is a violation of LRM:3.7(13).
> Both compilers will accept the code below.
> package foo is
>     type t is abstract tagged null record;
>     type i (f : access t'class) is abstract tagged limited null record;
>     type n is new t with null record;
>     type a_n is access all n'class;
>     type r (f : a_n) is new i (f) with null record;
> end foo;

Thanks for that. I don't quite see what (13) has to do with it?

I think I always get frightened by this sort of thing, cos clearly
(!!) the code I need (which is also acceptable to both compilers) is

package foo is
    type t is abstract tagged null record;
    type i (f : access t'class) is abstract tagged limited null record;
    type n is new t with null record;
    type r (f : access n'class) is new i (f) with null record;
end foo;

Referring to my original, Robert Dewar wrote:

> This is definitely invalid. GNAT says:
> 
> foo.ads:5:13: subtype must be compatible with parent
> discriminant

... so I won't need to report the fact that 3.11B _doesn't_ detect the
error, then!