Re: Proposal for communicator id

James Cownie (jim@meiko.co.uk)
Sun, 11 Jun 1995 12:11:25 +0000

> However, the first time MPI_COMM_ID() is called for a given
> communicator, the system is required to take the necessary steps
> to generate a communicator ID that satisfies the constraints
> imposed by point C.

So, it seems as if you're suggesting a function MPI_COMM_ID with the
following property :-

The FIRST time it is called on a communicator it is a collective
operation (and therefore may be a barrier), thereafter it is not.

This seems really horrible to me ! (Mostly because it is going to
introduce LOTS of bugs, especially when people use it on systems which
don't actually need the barrier, and then move to ones which do.) It
also seems unpleasant to have something which is there for profiling
purposes introduce such a large change in program behaviour as an
extra barrier...

-- Jim