Thanks for the quick response.
I just read formal semantics in the "Semantics for Multithread Java"
paper (ignoring the prescient stores and final variables). I'm also
pretty familiar with CRF, and had skimmed the "Java in CRF" paper,
but I missed the reference to children inheriting their parent's
state. Anyway, if I understand these correctly, Jan, et al.'s model
is much like release consistency, while Bill's is weaker, ensuring
synchronization only on a per-lock/volatile variable basis, similar
to a model I call "weak sequential locking". (Also a bit similar
to Iftode, et al.'s scope consistency, though that propagates even
less at synchronization points--and I think it guarantees a limited
form of coherence.)
In a few weeks, I'll try to think more about prescient writes
and final variables. (I'm in the midst of finishing up my thesis.)
But in the meantime, thanks again for the quick response, and the
pointers to the new semantics. That was very helpful.
JavaMemoryModel mailing list - http://www.cs.umd.edu/~pugh/java/memoryModel
This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:00:30 EDT