Re: JavaMemoryModel: Draft JMM Synopsis

From: Miles Sabin (miles@milessabin.com)
Date: Sun Mar 07 2004 - 12:30:15 EST


Jeremy Manson wrote,
> If you really wanted to change it, you could probably come up with
> some variant of "ordered before".

I guess Doug's objection to "happens before" is that as well as being
used in the spec as a very precisely defined technical term it also has
an informal meaning which could be misleading.

I'm sympathetic, but to be honest, I think it would be better to stick
to the same term otherwise there's a risk of a different kind of
confusion. If "happens before" is flagged up at the outset as being
used in the formal sense of the spec then that should reduce the risk
of people being mislead by their intuitions.

Still, I'd be happy with "ordered before" too, so long as there are
defining words up front to the effect of: i is ordered before j iff i
happens before j, where "happens before" is as defined in the spec.

Cheers,

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



This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:00:59 EDT