On May 25, 2004, at 5:07 PM, Bill Pugh wrote:
> Hans noticed that if we have an infinite sequence, then we may not be
> able to commit all of the actions in a finite number of set of
> committed actions,
> C1, ..., C_n.
>
> To fix this, we allow the sequence of executions used to show an
> execution is causal to be
> infinite. To make things even more complicated, we allow this sequence
> to have an ordinality greater
> than omega (since we also allow the same for actions).
>
We thought about this some more, and decided that the sequence of
commit steps
doesn't need to be able to have an ordinality greater than omega.
In other words, for any commit step C_i, there is only a finite number
of
commit step that come before it. However, we may, in fact, need an
infinite
number of commit steps.
Bill
-------------------------------
JavaMemoryModel mailing list - http://www.cs.umd.edu/~pugh/java/memoryModel
This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:01:08 EDT