From boo at terekhov.de  Wed Sep 15 14:11:40 2004
From: boo at terekhov.de (Alexander Terekhov)
Date: Wed, 15 Sep 2004 16:11:40 +0200
Subject: Memory model for multithreaded C++
Message-ID: <41484D9C.6BAA2770@terekhov.de>

< Forward Quoted >

-----
From: Ben Hutchings <ben-public-nospam at decadentplace.org.uk>
Newsgroups: comp.lang.c++.moderated
Subject: Re: Multithreaded programming: is the C++ standardization committee listening?
Date: 15 Sep 2004 05:57:02 -0400
Message-ID: <slrnckf9h5.pk.ben-public-nospam at decadentplace.org.uk>
References: ... <41475a8c at andromeda.datanet.hu>
-----

Ben Hutchings wrote:
> 
> Balog Pal wrote:
>  > "Joe Seigh" <jseigh_01 at xemaps.com> wrote in message
>  > news:41463A59.ABB040BD at xemaps.com...
>  >
>  >>    Formal definitions of the memory model were rejected as
>  >> unreadable by the vast majority of programmers.
>  >
>  > DOH
>  >
>  >> I could do a formal specification but I'm not really in a position
>  >> of any kind of authority so there's no point in my doing it, even
>  >> as an academic exercise.
>  >
>  > Of course there is.   C++ needs that formal spec. The to-be-assembled
>  > proposal on thread support must have it. (Hopefully the first move happens
>  > at the recent meeting.)
> 
> The first paper, entitled "Memory model for multithreaded C++", has
> been submitted as N1680.
> 
>  > For a while it may be just a TODO spot, but the sooner it is ready
>  > the better.
> 
> This first paper simply explains why the memory model needs to be
> redefined for multithreaded environments and what sort of special
> operations may be needed.
> 
>  > If you can do it please do, it will not be a wasted work.
> 
> It might be if it's duplicated work.  That's not meant to discourage
> anyone from working on the memory model; just to point out that people
> with experience are already doing so (Doug Lea and Bill Bugh were co-
> authors on both JSR 133 and this paper) and it might be a good idea to
> work with them than separately.
> 
> --
> Ben Hutchings
> If at first you don't succeed, you're doing about average.
> 
>       [ See http://www.gotw.ca/resources/clcm.htm for info about ]
>       [ comp.lang.c++.moderated.    First time posters: Do this! ]

regards,
alexander.


From boo at terekhov.de  Fri Sep 17 13:24:03 2004
From: boo at terekhov.de (Alexander Terekhov)
Date: Fri, 17 Sep 2004 15:24:03 +0200
Subject: [c++-pthreads] Memory model for multithreaded C++
References: <41484D9C.6BAA2770@terekhov.de>
Message-ID: <414AE573.7A926A03@terekhov.de>


< Forward Inline >

-------- Original Message --------
Newsgroups: comp.programming.threads
Subject: Re: C++ support for threads
Date: Fri, 17 Sep 2004 15:13:00 +0200
Message-ID: <414AE2DC.11DD838 at web.de>
References: <414ADC33.874AAFFE at xemaps.com>

Joe Seigh wrote:
> 
> It looks like there is some movement to support threads in C++ according
> to this link from a discussion on it in c.l.c++.m.
> 
> http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2004/n1680.pdf

They better stay out of volatile mess. atomic<> "intrinsic" is the
way to go. There's no need to "revise" C/C++ volatiles. Java's
volatiles imply more constraints than acquire-on-read and release-
on-write. BTW, Dinkum is pushing Boost.Thread (like) interface:

http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2004/n1682.html

They better first incorporate pthread.h as <cthread>...

http://groups.google.com/groups?selm=412533CA.82703A04%40web.de

is the next step.

regards,
alexander.