From: Niklas Holsti <niklas.holsti@tidorum.invalid>
Subject: Re: is there a version of unix written in Ada
Date: Sat, 3 Oct 2020 13:44:47 +0300 [thread overview]
Message-ID: <htr310Fh6phU1@mid.individual.net> (raw)
In-Reply-To: <rl9c0f$uf2$1@gioia.aioe.org>
On 2020-10-03 11:16, Dmitry A. Kazakov wrote:
> On 03/10/2020 09:44, Niklas Holsti wrote:
>> On 2020-10-03 9:42, Dmitry A. Kazakov wrote:
>>> On 03/10/2020 05:09, Randy Brukardt wrote:
>>>> "Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de> wrote in message
>>>> news:rl6its$10h7$1@gioia.aioe.org...
>>>
>>>>>> [If O3 was nested in O2, then the second call to O2 could be
>>>>>> internal.
>>>>>> But in that case, the first path would be impossible as O1 could
>>>>>> not see O3
>>>>>> to call it.]
>>>>>>
>>>>>> Remember that the decision as to whether a call is internal or
>>>>>> external
>>>>>> is
>>>>>> purely syntactic: if a protected object is given explicitly in the
>>>>>> call,
>>>>>> one
>>>>>> needs to trigger the mutual exclusion mechanisms again. The only
>>>>>> time one
>>>>>> doesn't need to do that is when the call does not include the object
>>>>>> (that
>>>>>> is, directly from the body of an operation).
>>>>>
>>>>> Even when the object in the call is statically known to be same?
>>>>
>>>> Yes. An external call *always* gets the lock again. I believe that
>>>> was made
>>>> the rule to make it obvious as to what will happen based on the form of
>>>> call.
>>>
>>> I mean this:
>>>
>>> protected body O is
>>> procedure P1 is
>>> begin
>>> ...
>>> end P1;
>>> procedure P2 is
>>> begin
>>> P1; -- OK
>>> O.P1; -- Deadlock or Program_Error
>>
>>
>> That is an internal call, so no deadlock nor error.
>>
>> See RM 9.5(4.e), which is this exact case.
>>
>>
>>> end P2;
>>> end O;
>
> I.e. it is *not* based on the syntax of the call.
At least not on /that/ syntactical difference.
> Anyway the rather disappointing result is that protected procedures may
> deadlock (or Program_Error) in a legal program.
Legal program can run into all sorts of problems, starting with
use-before-elaboration.
> So my initial disinclination to jump from one protected object to
> another is a reasonable advise.
Quite conservative advice, though.
> Or at least the order in which protected
> objects are navigated must be same.
I would say that it is advisable to arrange the POs (or PO types) in a
layered architecture and make inter-PO calls only from a higher-layer PO
to a lower-layer PO.
next prev parent reply other threads:[~2020-10-03 10:44 UTC|newest]
Thread overview: 105+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-10-01 22:21 is there a version of unix written in Ada Randy Brukardt
2020-10-02 6:55 ` Dmitry A. Kazakov
2020-10-03 3:09 ` Randy Brukardt
2020-10-03 6:42 ` Dmitry A. Kazakov
2020-10-03 7:44 ` Niklas Holsti
2020-10-03 8:16 ` Dmitry A. Kazakov
2020-10-03 10:44 ` Niklas Holsti [this message]
[not found] <e3c51e65-b71e-4623-bd58-c94cdbcc3ba6o@googlegroups.com>
2020-07-25 7:37 ` mockturtle
2020-07-26 19:45 ` Stéphane Rivière
2020-07-26 22:15 ` Jeffrey R. Carter
2020-07-27 7:40 ` Stéphane Rivière
2020-07-28 13:08 ` DrPi
2020-07-28 16:48 ` Simon Wright
2020-07-28 17:00 ` Fabien Chouteau
2020-07-29 9:21 ` DrPi
2020-07-29 9:28 ` Fabien Chouteau
2020-07-29 15:02 ` DrPi
2020-07-29 15:11 ` Fabien Chouteau
2020-07-29 20:41 ` DrPi
2020-07-29 9:20 ` DrPi
2020-07-29 16:53 ` Simon Wright
2020-07-29 20:42 ` DrPi
2020-07-25 8:47 ` Niklas Holsti
2020-07-25 9:36 ` Stéphane Rivière
2020-07-25 10:41 ` Luke A. Guest
2020-07-25 14:43 ` Jesper Quorning
2020-07-25 19:20 ` Andreas ZEURCHER
2020-07-27 7:40 ` Stéphane Rivière
2020-07-27 15:00 ` nobody in particular
2020-07-27 14:58 ` nobody in particular
2020-07-27 20:28 ` Shark8
2020-07-30 0:57 ` gdotone
2020-07-31 14:01 ` Shark8
2020-09-23 17:39 ` Olivier Henley
2020-09-25 15:06 ` DrPi
2020-09-25 17:31 ` Andreas ZEURCHER
2020-09-26 8:50 ` DrPi
2020-09-27 18:55 ` Luke A. Guest
2020-09-27 19:07 ` Luke A. Guest
2020-09-27 18:53 ` Luke A. Guest
2020-09-27 14:25 ` Shark8
2020-09-27 15:01 ` Dmitry A. Kazakov
2020-09-27 20:59 ` DrPi
2020-09-28 7:41 ` Dmitry A. Kazakov
2020-09-28 13:48 ` Olivier Henley
2020-09-28 14:48 ` Dmitry A. Kazakov
2020-09-28 16:28 ` Olivier Henley
2020-09-28 20:30 ` Dmitry A. Kazakov
2020-09-28 21:06 ` Shark8
2020-09-28 17:47 ` Paul Rubin
2020-09-28 20:27 ` Dmitry A. Kazakov
2020-09-29 23:54 ` Paul Rubin
2020-09-30 8:18 ` Dmitry A. Kazakov
2020-09-30 17:27 ` Paul Rubin
2020-09-30 19:42 ` Dmitry A. Kazakov
2020-09-30 20:33 ` Paul Rubin
2020-09-30 21:03 ` Dmitry A. Kazakov
2020-09-30 22:42 ` Randy Brukardt
2020-10-01 9:28 ` Dmitry A. Kazakov
2020-10-01 9:59 ` J-P. Rosen
2020-10-01 10:21 ` Dmitry A. Kazakov
2020-10-01 11:38 ` Niklas Holsti
2020-10-01 11:52 ` Niklas Holsti
2020-10-01 12:51 ` Dmitry A. Kazakov
2020-10-01 14:18 ` J-P. Rosen
2020-10-01 15:44 ` Dmitry A. Kazakov
2020-10-01 22:10 ` Randy Brukardt
2020-10-02 5:36 ` J-P. Rosen
2020-10-02 6:56 ` Dmitry A. Kazakov
2020-10-02 7:42 ` J-P. Rosen
2020-10-03 3:14 ` Randy Brukardt
2020-10-01 15:38 ` Niklas Holsti
2020-10-01 16:06 ` Dmitry A. Kazakov
2020-10-01 17:01 ` Niklas Holsti
2020-10-01 17:37 ` Dmitry A. Kazakov
2020-10-01 22:13 ` Randy Brukardt
2020-10-01 11:48 ` J-P. Rosen
2020-10-01 12:54 ` Dmitry A. Kazakov
2020-10-01 7:57 ` Paul Rubin
2020-10-01 9:26 ` Dmitry A. Kazakov
2020-10-01 9:46 ` Paul Rubin
2020-10-01 10:35 ` Dmitry A. Kazakov
2020-10-01 19:02 ` DrPi
2020-10-01 23:12 ` Paul Rubin
2020-10-01 21:54 ` Randy Brukardt
2020-10-01 23:14 ` Paul Rubin
2020-10-01 21:36 ` Brian Drummond
2020-10-02 6:56 ` Dmitry A. Kazakov
2020-10-02 18:34 ` Brian Drummond
2020-10-02 21:24 ` Paul Rubin
2020-10-06 23:54 ` Brian Drummond
2020-10-03 6:54 ` Dmitry A. Kazakov
2020-10-05 14:03 ` Shark8
2020-10-07 11:56 ` Olivier Henley
2020-10-09 12:35 ` Olivier Henley
2020-10-09 14:41 ` Brian Drummond
2020-09-28 18:40 ` Vincent Marciante
2020-09-28 20:28 ` Dmitry A. Kazakov
2020-09-28 17:04 ` DrPi
2020-09-28 17:30 ` Olivier Henley
2020-09-28 18:05 ` Olivier Henley
2020-09-28 19:36 ` Shark8
2020-09-03 10:32 ` c+
2020-09-12 4:30 ` sumde121
2020-09-19 14:09 ` erchetan33
replies disabled
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox