[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: [mpi-21] Call for MPI 2.2 and 3.0 agenda items for the Jan mee ting
- To: mpi-21@xxxxxxxxxxxxx
- Subject: RE: [mpi-21] Call for MPI 2.2 and 3.0 agenda items for the Jan mee ting
- From: "William Yu" <wyu@xxxxxxxxxx>
- Date: Sun, 2 Dec 2007 07:49:44 +0800
- Cc: mpi-21@xxxxxxxxxxxxx
- Delivered-to: mpifrm-mpi-21-outgoing@mailbouncer.mcs.anl.gov
- Delivered-to: mpifrm-mpi-21@mailbouncer.mcs.anl.gov
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:received:received:from:reply-to:to:cc:subject:date:message-id:x-mailer:mime-version:content-language:content-type:content-transfer-encoding:sender; bh=QfsmOIrLenVPbaXSkI17q4ZkiTMRAjnk64RR/aoDCgw=; b=Ym+Jy4GViHWXyW09S3OX9uWy2TK1wPTwN91KtKuLFvkhLBrUHPx72nqGWObmxmSHpCLlBfMOq9I6l4cAFyMJvxu0jZNgVs+g9hoXxJAo45ls9QvB+pLK3pdbej7SuudSaqnTqZJ2KsOS78pTbAqhKo/ZhZMRvC8NNt4sBsH8Iko=
- Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=received:from:reply-to:to:cc:subject:date:message-id:x-mailer:mime-version:content-language:content-type:content-transfer-encoding:sender; b=YJJ34dUbqHD4V9HTgfuJ6iWXHbGURTs91+wgUAwP3Y+GamsSxfb8vfr/bRkLgMM9KH7rE0pl3DhJXEvybBoh+egp6u8HUBcyl17h9nQ9Ni54RrUTPgKG9akDm6WLFHl17cmSNW/aZ9OEU7UxMh1lCA08NdttIJAZr3fcU9NUq0A=
- Reply-to: mpi-21@xxxxxxxxxxxxx
- Sender: owner-mpi-21@xxxxxxxxxxxxx
This definitely makes sense. You guys are probably right another forum is probably in a better position to define this.
Thanks!
________________ Reply Header ________________
Subject: RE: [mpi-21] Call for MPI 2.2 and 3.0 agenda items for the Jan meeting
Author: Erez Haba <erezh@xxxxxxxxxxxxx>
Date: December 01st 2007 9:08 pm
I agree with Jeff here, the higher abstraction languages require more than a plain down level interface like C/FORTRAN. To make them usable, a more abstract programming model is needed. One that fits naturally with the language and its usage. For example, see the excellent bindings for C# with MPI.NET from the University of Indiana http://www.osl.iu.edu/research/mpi.net/.
If you look at this interface you'll find that many of its aspects are not even close to the way the existing MPI interfaces are defined. These differences have very good reason, they are to give better support for the language abstractions and be natural for the C# users.
For example, the entire MPI_Type* set of API's are not present in this interface as they've been abstracted away by the MPI.NET C# bindings. This enables a more natural use for the C# programmer like auto type-safety marshaling of any data type or calling collectives with a variable length string.
I think that this level of interface is beyond the scope of the MPI forum and is more in the scope of the specific language experts.
Thanks,
.Erez
-----Original Message-----
From: owner-mpi-21@xxxxxxxxxxxxx [mailto:owner-mpi-21@xxxxxxxxxxxxx] On Behalf Of Jeff Squyres
Sent: Saturday, December 01, 2007 8:55 AM
To: mpi-21@xxxxxxxxxxxxx
Subject: Re: [mpi-21] Call for MPI 2.2 and 3.0 agenda items for the Jan mee ting
On Nov 30, 2007, at 10:44 PM, William Emmanuel S. Yu wrote:
> That is right Jeff. Official Java Bindings would be a good first step.
> The goal is to have "consistent" guidelines (standards) for library
> makers.
>
> What is the best way to go about this?
All new MPI topics start with a proposal. :-)
But before you spend time on a proposal, you might want to ask the
community if they want official Java bindings. There were strong
waves of apathy to prior attempts at standardizing Java bindings.
> The Boost C++ implementation is
> pretty good but only works on ... well ... Boost. No plans to define
> something more consistent across the board? Or maybe this should be
> done by another group?
I'm not sure what you're asking here. The Boost MPI package is a fine
standalone package. I don't see any reason to standardize it; MPI
already has basic-building-blocks C++ bindings.
--
Jeff Squyres
Cisco Systems