[mpich-devel] MPIR_Comm_split_type_impl

[Pavan Balaji]

Good catch.  Can you file a ticket for this?

 -- Pavan

On 11/15/2012 01:50 PM US Central Time, Jeff Hammond wrote:
> I haven't confirmed this yet, but I'm not filing a bug because perhaps
> this instead should be the subject of an MPI-3 erratum discussion.
> 
> For those who - like me - (didn't) don't know what "horks" mean, I
> recommend that you do not Google it.  The definition from
> urbandictionary.com is quite foul.
> 
> Best,
> 
> Jeff
> 
> 
> ---------- Forwarded message ----------
> From: Michael Raymond <mraymond at sgi.com>
> Date: Thu, Nov 15, 2012 at 1:15 PM
> Subject: Re: [Mpi-forum] MPI 3.0 implementation slides
> To: Jeff Hammond <jhammond at alcf.anl.gov>
> 
> 
>   As a way of saying thank you for all the great info:
> 
> MPIR_Comm_split_type_impl() horks on split_type == MPI_UNDEFINED even
> though the 3.0 standard allows it.
> 
> 
> --
> Michael A. Raymond
> SGI MPT Team Leader
> (651) 683-3434
> 
> 
> 

-- 
Pavan Balaji
http://www.mcs.anl.gov/~balaji