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=-1.9 required=5.0 tests=BAYES_00
	autolearn=unavailable autolearn_force=no version=3.4.4
Path: 
 border1.nntp.dca3.giganews.com!backlog3.nntp.dca3.giganews.com!border1.nntp.dca.giganews.com!nntp.giganews.com!goblin2!goblin.stu.neva.ru!reality.xs3.de!news.jacob-sparre.dk!loke.jacob-sparre.dk!pnx.dk!.POSTED!not-for-mail
From: "Randy Brukardt" <randy@rrsoftware.com>
Newsgroups: comp.lang.ada
Subject: Re: Oberon and Wirthian languages
Date: Tue, 22 Apr 2014 18:30:38 -0500
Organization: Jacob Sparre Andersen Research & Innovation
Message-ID: <lj6u2u$7a9$1@loke.gir.dk>
References: <1ljwj8f.1wqbhvuabsdw1N%csampson@inetworld.net>
 <51c7d6d4-e3be-44d5-a4ce-f7e875345588@googlegroups.com>
 <li9g34$9q8$1@dont-email.me> <%J32v.70539$kp1.45343@fx14.iad>
 <lic17o$re4$1@dont-email.me> <8761m535e4.fsf_-_@ludovic-brenta.org>
 <lj3540$m81$1@dont-email.me> <lj3nao$u22$2@dont-email.me>
 <lj4de1$jql$1@dont-email.me>
NNTP-Posting-Host: static-69-95-181-76.mad.choiceone.net
X-Trace: loke.gir.dk 1398209439 7497 69.95.181.76 (22 Apr 2014 23:30:39 GMT)
X-Complaints-To: news@jacob-sparre.dk
NNTP-Posting-Date: Tue, 22 Apr 2014 23:30:39 +0000 (UTC)
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2900.5931
X-RFC2646: Format=Flowed; Original
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.6157
X-Original-Bytes: 2692
Xref: number.nntp.dca.giganews.com comp.lang.ada:185949
Date: 2014-04-22T18:30:38-05:00
List-Id: <comp.lang.ada>

"Simon Clubley" <clubley@remove_me.eisner.decus.org-Earth.UFP> wrote in 
message news:lj4de1$jql$1@dont-email.me...
...
>> If the compiler accepts the pragma, then assignment to Register (as 
>> opposed to
>> its components) is atomic:
>>
>> Register := (A => True, ...);
>
> Thank you. It was not at all 100% clear from the RM if Ada offered the
> _guarantee_ that if you had, say, 3 components in a record assignment
> for a atomic bitfield record, that the record assignment would be treated
> as one atomic operation instead of 3 atomic operations (one for each
> component).

It's not quite a guarentee, as it's Implementation Advice (C.6(22-23/2). The 
reason for it being IA is simply that we can't use undefined terms like 
"load" and "store" in normative wording. In this case, IA is actually 
stronger because we can say directly what we mean without defining it 
exactly. In theory, implementations are supposed to document all cases where 
they do not follow Implementation Advice, so you should be able to determine 
for a particular implementation as to whether it is followed.

I don't think there could be an absolute guarantee, as the idea of a load or 
store might not make sense on a data-flow machine (or some other exotic 
design). I would guess that implementers do this in virtually all cases. As 
Jeff said, if it is really critical, you have a verify the code.

                                    Randy.