RE: JavaMemoryModel: The property of "race-free implies SC"

From: Sarita Adve (sadve@cs.uiuc.edu)
Date: Fri Oct 31 2003 - 17:07:21 EST


> Therefore, in my opinion, we may indeed specify JMM to be that non-SC
> model that we want to use when reasoning about race-free'ness.
>
> I'll be glad if you can further convince me :)

It's just too hard for (1) most programmers to deal with anything other than
SC, and (2) for people like us to agree on anything other than SC! From my
last message:

Now you could argue that we should really do away with this SC stuff. But
pragmatically speaking, any other model will be a lot more complex, and for
any model you come up with, others will come up with something weaker or
stronger that makes more sense to them. That is, we will forever have
languages and systems with different models creating a
programming/portability nightmare.

Sarita

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



This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:00:52 EDT