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=0.7 required=5.0 tests=BAYES_00,INVALID_DATE,
	REPLYTO_WITHOUT_TO_CC autolearn=no autolearn_force=no version=3.4.4
Xref: utzoo comp.sys.encore:560 comp.lang.ada:3557
Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!yale!husc6!encore!jcallen
From: jcallen@Encore.COM (Jerry Callen)
Newsgroups: comp.sys.encore,comp.lang.ada
Subject: Re: Strange behavior by Ada tasks
Summary: Strange, but legal
Message-ID: <11444@encore.Encore.COM>
Date: 27 Mar 90 21:46:46 GMT
Reply-To: jcallen@encore.UUCP (Jerry Callen)
Followup-To: comp.sys.encore
Organization: Encore Computer Corp, Marlboro, MA
List-Id: <comp.lang.ada>


kpierce@umn-d-ub.D.UMN.EDU (Keith Pierce) writes:

 [Introductory comments regarding use of Ada tasking to implement
  semaphores and protected variables a la Booch deleted.]


>with Semaphore;
>
>package Nat_Package is
>  type Protected_Natural is limited private;
>  procedure Add_One (To_Number : in out Protected_Natural);
>  procedure Put (The_Number : in Protected_Natural);
>private
>  type Protected_Natural is
>    record
>      The_Value : Natural := 0;
>      Guard : Semaphore.Kind;
>    end record;
>end Nat_Package;

 [More code and stuff deleted]

The problem lies in the fact that Protected_Natural is being passed to
Add_One by *copy* and not by *reference*. This is legal (see LRM 6.2, 
paragraphs 6 through 8), but it means that the two tasks are updating
DIFFERENT copies of The_Value (each task has a copy on its own stack).

I was able to make your code work by making Protected_Natural an access
type that points to the record containing the natural and the task and
adding a Create procedure to allocate the record. This causes both tasks
to access the same (the only) copy of The_Value.


-- Jerry Callen
   jcallen@encore.com
   (508) 460-0500 (work)