[concurrency-interest] question the about the JMM

Larry Riedel larryr at saturn.sdsu.edu
Fri Dec 7 01:04:35 EST 2007


> > I presumed a "logical model for the JMM" would be addressing
> > propagation of object state changes between threads.  The
> > thread above does not seem to be interested in making sure
> > it has the latest state information for any objects...
> 
> The model explains *why* the code shown does not ensure the
> latest state information is communicated. The thread *is*
> interested, but the code is wrong under the JMM.

I am inclined to think the way to make it not wrong is to use
mechanisms which introduce something which can be described
in terms of a "happens-before" model, or in terms of a cache
coherent distributed shared memory model, with comparable
levels of intuitiveness and accuracy.  I am not sure though.


Larry



More information about the Concurrency-interest mailing list