Re: JavaMemoryModel: How bad are incorrectly synchronized programs?

From: Assaf Schuster (assaf@cs.technion.ac.il)
Date: Fri Aug 01 2003 - 07:29:23 EDT


----- Original Message -----
From: <victor.luchangco@sun.com>
To: <javamemorymodel@cs.umd.edu>
Sent: Thursday, July 31, 2003 10:18 AM
Subject: JavaMemoryModel: "Causality"

> I believe all the currently proposed models are intended to forbid
> both causal loops and phantom writes, though Sarita argues that her
> "smoking gun" example shows that CnC (and therefore all the proposed
> models, except possibly the SC- formulation of CnC) allow some form of
> causal loop.

I do not understand this par.
As far as I can see, SC-1 allows the smoking gun too (does it claim not
to?).
(perhaps this is consistent with US perception of smoking guns:
  if *we* hold the gun then it is not smoking :)

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



This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:00:49 EDT