Proposed rephrasing of the restriction on BES vector operations in the HPC Profile

Hi; As you may recall from the discussions at the Tokyo GGF conference, it was agreed that the HPC Profile base case would be kept as simple as possible in order to make it as easy as possible to create a compliant BES service (and to forestall feature creep). Among other things, that meant that we agreed upon requiring the HPC Profile-compliant BES services would only be required to support BES vector operations of length 1. But of course we'd like to allow compliant BES services to support vectors of length greater than 1 if they so desire (which we expect to be the common case). Enclosed is a rewording of the paragraph in section 4 of the HPC Profile specification draft that restricts vector lengths of BES operations. I've rewritten it to clarify that vector lengths greater than 1 are allowed but may result in a fault. Please let me know if you would like additional rewording to occur. If I don't hear any rewording suggestions in the next few days then I'll go ahead and insert the rewording shown into the HPC Profile document. Reworded paragraph: The BES GetActivitiesStatus, TerminateActivities, and GetJSDLDocuments operations include a vector input parameter that specifies the set of activities that the operation should be applied to. A Compliant system MUST be prepared to support a vector length of 1. A Compliant system SHOULD support input vector lengths greater than 1 but MAY return a Not_Supported fault in response to input vector lengths greater than 1. Thanks, Marvin.

Sounds good. But I would like to see an albeit-trivial change: OLD: "A Compliant system MUST be prepared to support a vector length of 1." NEW: ""A Compliant system MUST support a vector length of 1." I told you it was trivial, right? :^) - Marty Quoting Marvin Theimer <theimer@microsoft.com>:
Hi;
As you may recall from the discussions at the Tokyo GGF conference, it was agreed that the HPC Profile base case would be kept as simple as possible in order to make it as easy as possible to create a compliant BES service (and to forestall feature creep). Among other things, that meant that we agreed upon requiring the HPC Profile-compliant BES services would only be required to support BES vector operations of length 1. But of course we'd like to allow compliant BES services to support vectors of length greater than 1 if they so desire (which we expect to be the common case).
Enclosed is a rewording of the paragraph in section 4 of the HPC Profile specification draft that restricts vector lengths of BES operations. I've rewritten it to clarify that vector lengths greater than 1 are allowed but may result in a fault. Please let me know if you would like additional rewording to occur. If I don't hear any rewording suggestions in the next few days then I'll go ahead and insert the rewording shown into the HPC Profile document.
Reworded paragraph: The BES GetActivitiesStatus, TerminateActivities, and GetJSDLDocuments operations include a vector input parameter that specifies the set of activities that the operation should be applied to. A Compliant system MUST be prepared to support a vector length of 1. A Compliant system SHOULD support input vector lengths greater than 1 but MAY return a Not_Supported fault in response to input vector lengths greater than 1.
Thanks, Marvin.
---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program.

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Marvin, all, Marvin Theimer wrote: | [...] | | Reworded paragraph: | | The BES GetActivitiesStatus, TerminateActivities, and GetJSDLDocuments | operations include a vector input parameter that specifies the set of | activities that the operation should be applied to. A Compliant system | MUST be prepared to support a vector length of 1. A Compliant system | SHOULD support input vector lengths greater than 1 but MAY return a | Not_Supported fault in response to input vector lengths greater than 1. This sounds good. However, on the phone conference on Friday, we had a a longer discussion on the return vector sizes. Should we emphasize that if a compliant service supports input vectors longer than 1 that this service must return vectors of the same length as the input vector? Leaving this out creates a gut feeling of "something's missing here" for me. Cheers, Michel - -- Michel <dot> Drescher <at> uk <dot> fujitsu <dot> com Fujitsu Laboratories of Europe +44 20 8606 4834 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.3 (Darwin) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFE8U7Rk0lMZTNKw4QRAikWAJ94G5WJ0akTUfFlzSnM+Eek3E96XwCfeOe9 dT18+7WKdNGsxRAg8ALP03c= =A8YN -----END PGP SIGNATURE-----

Michel Drescher wrote:
Should we emphasize that if a compliant service supports input vectors longer than 1 that this service must return vectors of the same length as the input vector?
Leaving this out creates a gut feeling of "something's missing here" for me.
To me, the required matching of input and output vector lengths would seem to be something that ought to be required in the BES spec, and not the profile of it. I believe it is something that ought to be true of all BES implementations, yes? Donal.

Hi; I agree with you on this. The BES specification is where this topic needs to be explicated rather than the HPC Profile. It's something I intend to bring up at the next telecon call. Marvin. -----Original Message----- From: owner-ogsa-hpcp-wg@ggf.org [mailto:owner-ogsa-hpcp-wg@ggf.org] On Behalf Of Donal K. Fellows Sent: Tuesday, August 29, 2006 12:58 AM To: Michel Drescher Cc: ogsa-hpcp-wg@ggf.org Subject: Re: [ogsa-hpcp-wg] Proposed rephrasing of the restriction on BES vector operations in the HPC Profile Michel Drescher wrote:
Should we emphasize that if a compliant service supports input vectors longer than 1 that this service must return vectors of the same length as the input vector?
Leaving this out creates a gut feeling of "something's missing here" for me.
To me, the required matching of input and output vector lengths would seem to be something that ought to be required in the BES spec, and not the profile of it. I believe it is something that ought to be true of all BES implementations, yes? Donal.

Marvin Theimer wrote:
Hi;
As you may recall from the discussions at the Tokyo GGF conference, it was agreed that the HPC Profile base case would be kept as simple as possible in order to make it as easy as possible to create a compliant BES service (and to forestall feature creep). Among other things, that meant that we agreed upon requiring the HPC Profile-compliant BES services would only be required to support BES vector operations of length 1. But of course we’d like to /allow/ compliant BES services to support vectors of length greater than 1 if they so desire (which we expect to be the common case).
Enclosed is a rewording of the paragraph in section 4 of the HPC Profile specification draft that restricts vector lengths of BES operations. I’ve rewritten it to clarify that vector lengths greater than 1 are allowed but may result in a fault. Please let me know if you would like additional rewording to occur. If I don’t hear any rewording suggestions in the next few days then I’ll go ahead and insert the rewording shown into the HPC Profile document.
Reworded paragraph:
The BES GetActivitiesStatus, TerminateActivities, and GetJSDLDocuments operations include a vector input parameter that specifies the set of activities that the operation should be applied to. A Compliant system MUST be prepared to support a vector length of 1. A Compliant system SHOULD support input vector lengths greater than 1 but MAY return a Not_Supported fault in response to input vector lengths greater than 1.
One minor correction: the latest BES spec replaces "GetActivitiesStatus" with "GetActivityStatuses" and "GetJSDLDocuments" with "GetActivityDocuments". Peter
Thanks,
Marvin.
participants (5)
-
Donal K. Fellows
-
humphrey@cs.virginia.edu
-
Marvin Theimer
-
Michel Drescher
-
Peter G. Lane