Bill Pugh wrote:
> The weak vs. strong interpretation only involves the effects of
> volatiles reads and writes on non volatile reads and writes.
But that would seem to be the crux of using volatiles as a means to
synchronize access to non-volatiles in lock-free situations. ???
This distinction between the weak and strong interpretation is too subtle
for me to grasp the implications of it for the end user.
But my gut feeling is that the weak interpretation is going to require that
lock-free algorithms are analysed in far more depth than would be necessary
under the stronger interpretation.
David Holmes
-------------------------------
JavaMemoryModel mailing list - http://www.cs.umd.edu/~pugh/java/memoryModel
This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:01:00 EDT