George and I talked on the phone about this. See:
On Aug 4, 2008, at 9:55 AM, Jeff Squyres wrote:
> On Aug 4, 2008, at 9:52 AM, Brian Barrett wrote:
>> I suppose, but that's a pain when you just want to say "I don't
>> support calling add_procs a second time" :). But I'm not going to
>> fix all the BTLs to make that work right, so I suppose in the end I
>> really don't have a strong opinion.
> Good point. I think that if the PML/BML sees OMPI_ERR_NOT_SUPPORTED
> or OMPI_ERR_NOT_IMPLEMENTED from add_procs(), that can be non-fatal;
> the BTL can just be ignored for this round (i.e., it wouldn't affect
> any procs that were already successfully added to that BTL module).
> Of course, that means that a BTL could return NOT_SUPPORTED/
> IMPLEMENTED during MPI_INIT, but that would be pretty dumb. :-)
> Jeff Squyres
> Cisco Systems
> devel mailing list