I just wanted to affirm what Bill said, and also to
retract some of what I said in my previous email about
liveness. The liveness condition I stated was too
strong. It should have just been that in a finite
execution, every thread terminates (no requirement on
infinite executions). I'm still ambivalent about
providing any liveness conditions in the JMM itself
(see my previous email), but this one is quite weak,
which is good.
As Bill said, I think the formulation would be better
using ordinary internal actions, but this is just a
matter of presentation, not of semantics, which is
what we really need to agree on. I haven't read the
latest actual document carefully on this point, but
I will do so.
Victor
-------------------------------
JavaMemoryModel mailing list - http://www.cs.umd.edu/~pugh/java/memoryModel
This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:01:07 EDT