KAI C++ statics

[Andrei Alexandrescu]

I allowed myself to pass our doc to Arch Robison (one of the authors of 
KAI C++) and he commented on their implementation of static, see below.

Andrei

-------- Original Message --------
From: Robison, Arch <arch.robison at intel.com>
Subject: RE: Memory model for multithreaded C++
To: Andrei Alexandrescu <andrei at metalanguage.com>

Thanks.  With regard to function-scope statics, KAI C++ added the
synchronization code.  No one complained about the performance of it,
even though it was a fairly heavy-weight implementation for the
initialization.  It was based on double-check, which *does* work when
you own the compiler :-)  It was not a big deal to add the
synchronization code.

- Arch

-----Original Message-----
From: Andrei Alexandrescu [mailto:andrei at metalanguage.com]
Sent: Tuesday, March 08, 2005 1:08 PM
To: Robison, Arch
Subject: Re: Memory model for multithreaded C++

Actually, yes. We've just submitted something, see attached.

Andrei

Robison, Arch wrote:
 > I read the Sept. 2004 document with interest.  Has there been any
 > further effort on fleshing out the API?
 >
 > Arch D. Robison          Intel Corporation
 > arch.robison at intel.com   Parallel & Distributed Solutions Division
 > (217)-403-4240           1906 Fox Drive
 >                          Champaign IL 61820
 >
 >
 >