Wed Oct 14 14:17:13 EDT 2015
Bill Long wrote:
> See 15-244r1.
I don't think this is quite right yet. It is the count that is accessed
atomically and it is also altered by EVENT WAIT. I suggest
"The count of the EVENT argument is accessed atomically with respect to
the execution of EVENT POST and EVENT WAIT statements in unordered
segments, in exact analogy to the ATOM argument of an atomic subroutine."
BTW, if an error occurs in an EVENT POST or EVENT WAIT statement, the
value of the count is processor dependent, so we are not consistent in
the TS in giving it a value if EVENT_QUERY fails.
More information about the J3