In the parallel extension we have created in GridSAM. We have a new element called <mpi:MPIApplication/> that extends the jsdl:posix:POSIXApplication_Type. Therefore the MPIApplication inherits the arguments, executable, etc. with the additional elements to specify parallelism. Roughly.... <xsd:complexType name="MPIApplicationType"> <xsd:complexContent> <xsd:extension base="jsdl-posix:POSIXApplication_Type"> <xsd:sequence> <xsd:element ref="ProcessorCount"/> </xsd:sequence> </xsd:extension> </xsd:complexContent> </xsd:complexType> <xsd:element name="ProcessorCount" type="xsd:positiveInteger"/> <xsd:element name="MPIApplication" type="MPIApplicationType"/> William --- William Lee - Software Coordinator --- --- London e-Science Centre, Imperial College London --- A: Room 380, Department of Computing, Imperial College London, Huxley Building, South Kensington campus, London SW7 2AZ, UK E: wwhl at doc.ic.ac.uk | william at imageunion.com W: www.lesc.ic.ac.uk | www.imageunion.com P: +44(0)20 7594 8310 F: +44(0)20 7581 8024 -----Original Message----- From: owner-jsdl-wg@ggf.org [mailto:owner-jsdl-wg@ggf.org] On Behalf Of Donal K. Fellows Sent: 23 February 2006 09:11 To: JSDL WG Subject: Re: [jsdl-wg] Future plans for JSDL A S McGough wrote:
3) Prioritise the items. From the session parallel jobs seemed to be the thing most people wanted next.
Just a quick note to say that the main reason for doing the parallel jobs stuff next (probably starting from the base proposed by the people from Barcelona; they've identified the stuff that's needed after all) is that otherwise there'll be several different groups doing it themselves in an incompatible way. To me, that says it is an *excellent* idea to foster a standard extension for this and to expedite this. OK, I think other extensions/improvements are more important in the grand scheme of things, but the fact that people are doing it now forces our hand. Looking at the parallel proposal, I have to ask how it is supposed to be combined with the POSIX extension, since without that it's not going to work (still need command-line args, env-vars, etc.) Once we've nailed that down, I think it will be just A Simple Matter Of Writing. :-) Donal.