Re: JavaMemoryModel: IMPORTANT: New Unified JMM formalism, $100 reward for any flaws

From: Bill Pugh (pugh@cs.umd.edu)
Date: Wed Mar 17 2004 - 12:22:54 EST


On Mar 16, 2004, at 9:22 PM, Sylvia Else wrote:

> At 08:01 PM 16/03/2004 -0500, Bill Pugh wrote:
>
>
> We put U7 in the document because we weren't thrilled with it.
>
> It is pretty horrible, but after all the code is not correctly
> synchronized. It did though raise another question in my mind. Is it
> true that all correctly synchronized code fragments will show SC
> behaviour?
>
> Sylvia
>

That depends on what your definition of "fragment" is.

If a fragment is a set of threads that definitely has no interaction
with any other threads, that this property is clearly so.

I suspect we have the Isolation principle I postulated before, but
we haven't done a proof of it.

        Bill

-------------------------------
JavaMemoryModel mailing list - http://www.cs.umd.edu/~pugh/java/memoryModel



This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:00:59 EDT