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
Path: eternal-september.org!reader02.eternal-september.org!aioe.org!5WHqCw2XxjHb2npjM9GYbw.user.gioia.aioe.org.POSTED!not-for-mail
From: "Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de>
Newsgroups: comp.lang.ada
Subject: Re: ANN: GCC 11.1.0 for macOS
Date: Tue, 4 May 2021 22:12:42 +0200
Organization: Aioe.org NNTP Server
Message-ID: <s6s9rm$1clb$1@gioia.aioe.org>
References: <lyeeep2j8m.fsf@pushface.org> <s6mlfu$eii$1@gioia.aioe.org>
 <86v97yie5y.fsf@stephe-leake.org>
NNTP-Posting-Host: 5WHqCw2XxjHb2npjM9GYbw.user.gioia.aioe.org
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 7bit
X-Complaints-To: abuse@aioe.org
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:78.0) Gecko/20100101
 Thunderbird/78.10.0
X-Notice: Filtered by postfilter v. 0.9.2
Content-Language: en-US
Xref: reader02.eternal-september.org comp.lang.ada:61953
List-Id: <comp.lang.ada>

On 2021-05-04 19:47, Stephen Leake wrote:
> "Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de> writes:
> 
>> As a side note. The version 11 brings new incompatibilities breaking
>> old code. In some cases X'Access is no more accepted and need to be
>> replaced by X'Unchecked_Access.
>>
>> I am to lazy to analyze whether that it is a bug of feature, just be aware.
>>
>> I dare say that every Ada style guideline should require
>> 'Unchecked_Access everywhere. The issue became a permanent maintenance
>> nightmare.
> 
> I had a similar issue upgrading from GNAT Community 2020 to GNAT Pro 21.
> The GNAT compiler has gotten smarter about enforcing accessibility
> rules.

Alas.

> Since those rules are there to prevent dangling references, they should
> be respected; I fixed my code to compile with 'Access.

For example:

    declare
       Location : Abstract_Layer'Class renames
                  Abstract_Layer'Class (Under.all);
    begin
       ...
       if Location'Access = Location.Widget.Bottom then

This does not compile anymore. Clearly there cannot be any dangling 
references here.

Recent changes broke a lot of code involving comparisons of access 
types, especially if an anonymous access type is involved. Among them 
are cases when even 'Unchecked_Access does not help. So, one should 
resort to awful 'Address instead.

The situation is quite dire. I would even suggest introducing a built-in 
operation to compare an object with an access, e.g.

    P'Refers (X)  -- True if P points to X

since comparison of access types became too volatile.

> It is a pain that GNAT didn't get this totally correct the first time
> around, but that's life.

I am not a language lawyer to judge. My impression that in practice 
accessibility rules significantly reduce safety and code quality rather 
than add it.

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de