>MPI_WIN_BARRIER should imply MPI_BARRIER
NO; we discussed this at length, and the decision was for weaker
synchronization. All the RMA synchronization calls are synchronizing RMA
operations, but avoid to promise more.
Then, PLEASE lets NOT call it MPI_WIN_BARRIER -- to have a function called
barrier, which doesn't do a barrier is just too confusing. What about
MPI_WIN_SYNC.
>clarification on what can block in a START, PUT, COMPLETE sequence.
OK, will try to improve prose. Jusdt to make sure people understand the
current text: I assume that the following code may not deadlock
process 0 process 1
start post
put
send ---> recv
complete wait
I assume you mean "will not deadlock"
My point is that the following code MAY not deadlock (some implementations it
will, some it wont):
process 0 process 1
start
put
complete
send ---> recv
post
wait
The current text might imply that (al least) the complete is synchronized with
the post.
Lloyd Lewins
Hughes Aircraft Co.,
llewins@msmail4.hac.com