Thanks for that.
I see that rule 6 is a useful one to get things started, but in itself it
doesn't seem to prevent a commit of a write which does not happen-before a
read (as in the example of Figures 10 and 11, page 22 of the spec), so
committing x = 1 in C1 in the Fig 6 example seemed no different from
committing y = 1 in that example.
Still not seeing why the rules of section 7.4 in themselves take account of
issues like conditionals; perhaps what I'm missing is how the intra-thread
semantics limit possible well-formed executions. Time to think more!
It's fast, it's easy and it's free. Get MSN Messenger today!
JavaMemoryModel mailing list - http://www.cs.umd.edu/~pugh/java/memoryModel
This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:01:09 EDT