On Feb 26, 2004, at 12:00 PM, Vijay Saraswat wrote:
> Slippery slope, and nothing to stop you.
> "false" is just a shorthand for something that may be too complicated
> to figure out is false.
> One of the things that is worrisome is that introducing this extra
> "irrelevant" code *prohibits* a behavior. That seems just wrong.
I don't believe the model should prohibit this behavior. With thread 3
in place (i.e., case 18a) the model should still allow r1 == r2 == r3
Introducing additional irrelevant code may mean that a particular
compiler transformation won't allow the behavior. But that doesn't mean
that the behavior can't be or isn't allowed.
JavaMemoryModel mailing list - http://www.cs.umd.edu/~pugh/java/memoryModel
This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:00:58 EDT