Research Group / Working Groups
Hi, We discussed the research group/working group structure on today's call, present were myself, Andre, Shantenu, Thilo, Pascal and Hartmut. The consensus was to go down the umbrella research group with spawned working group road. To be concrete: We propose to split the current RG in two: SAGA-RG: This research group will be responsible for deciding look and feel of the API, identifying new SAGA subsystems which we should/might want to have APIs for, spawning working groups to look into them (e.g. after a design team has come up with a straw-man), and coordinating the resulting working groups. It could look into issues with OGSA alignment, or work with OGSA to spawn a group to look into common issues. This group will inherit the current charter, but remove the API document deliverables and add some text describing the new scope, the process for spawning working groups, and the relationship between the RG and the groups. Proposed chairs: Tom Goodale, Shantenu Jha, Thilo Kielmann SAGA-CORE-WG This group will concentrate on producing an API document from the current strawman. It will inherit the current charter, but removing the use case and requirements document deliverables. The charter will be refined to specify precisely the areas covered by the strawman as the scope of the WG, and will define how it relates to the RG. The timeline for producing the API documents will be unchanged. Proposed chairs: Tom Goodale, Andre Merzky Note that the chairs proposed above are just suggestions, and will need to be ratified by the group. If anyone else would like to become a chair, or if anyone has an issue with the proposed chairs, please don't hesitate to speak. This reorganisation will hopefully produce a clear delineation of the roles of the research group and the working group, and provide a mechanism for us to spawn more working groups to look at other subsystems such as GridCPR and GridRPC, or start SAGA activities within such groups if that would be appropriate. Creating new WGs in this way would, we hope, make it easier for people to engage in the process of defining new APIs, and provide a much clearer process for the generation of these APIs. We do have some worries, though, which we need to discuss before finalising on this route: 1) Will the additional admin overhead be worth the gain ? The feeling on the call was that this approach has potential to get more people involved, and worst case, leaves us in the current situation with an active WG and an inactive RG. 2) Will this approach really help us to engage and attract new people ? We are hoping that the ability to spawn small, tightly-focussed groups will help make it easier to attract people and focus them on the API development. We are planning to have another conference call next Wednesday at 1400 GMT (same time as today's call), to continue discussion of this, and on Thursday wish to send the decision of the group to the GFSG so that they may discuss it before GGF. I know this is a short time-frame, but we need to finalise this issue soon, and comments from the wider group are essential. Please speak up one way or another in the next few days as to whether we should go forward with this plan, or ask the GFSG to just 'flip-the-bit' as per the original discussions. Any other comments or suggestions would be great. Even a response such as 'no, just flip the bit', or 'go for it' would be helpful. We have assurance from the ADs that whichever way we decide to go, the transition will be quick and relatively painless. Cheers, Tom
Hi all, "the most useless part of the discussion takes up 90 percent of the time. Minimum." (Andres Leaden Rule) I think the name SAGA-CORE-WG is misleading, as the strawman does not really define a Core API. That role, defining the core, look and feel, and basic paradigms, would rather fit the target of the RG. However, during yesterdays discussion we could not come up with a better name, really. Just so you all have something to laugh at, I'll repeat some suggestions below. Well, the point of this mail is: if you can think of a suitable group name, please speak up! iSAGA (very fashionable, for initial SAGA) SAGA-PRIMER (obvious) SSAGA (confusing, but stands for Super simple API...) My personal favourite for today is (cling on tight) SAGA-WG Well, that might be confusing with SAGA-RG, but what the heck - it describes what we do best I think... Cheers, Andre. Quoting [Tom Goodale] (Feb 04 2006):
Hi,
We discussed the research group/working group structure on today's call, present were myself, Andre, Shantenu, Thilo, Pascal and Hartmut.
The consensus was to go down the umbrella research group with spawned working group road.
To be concrete:
We propose to split the current RG in two:
SAGA-RG:
This research group will be responsible for deciding look and feel of the API, identifying new SAGA subsystems which we should/might want to have APIs for, spawning working groups to look into them (e.g. after a design team has come up with a straw-man), and coordinating the resulting working groups. It could look into issues with OGSA alignment, or work with OGSA to spawn a group to look into common issues.
This group will inherit the current charter, but remove the API document deliverables and add some text describing the new scope, the process for spawning working groups, and the relationship between the RG and the groups.
Proposed chairs: Tom Goodale, Shantenu Jha, Thilo Kielmann
SAGA-CORE-WG
This group will concentrate on producing an API document from the current strawman. It will inherit the current charter, but removing the use case and requirements document deliverables. The charter will be refined to specify precisely the areas covered by the strawman as the scope of the WG, and will define how it relates to the RG. The timeline for producing the API documents will be unchanged.
Proposed chairs: Tom Goodale, Andre Merzky
Note that the chairs proposed above are just suggestions, and will need to be ratified by the group. If anyone else would like to become a chair, or if anyone has an issue with the proposed chairs, please don't hesitate to speak.
This reorganisation will hopefully produce a clear delineation of the roles of the research group and the working group, and provide a mechanism for us to spawn more working groups to look at other subsystems such as GridCPR and GridRPC, or start SAGA activities within such groups if that would be appropriate. Creating new WGs in this way would, we hope, make it easier for people to engage in the process of defining new APIs, and provide a much clearer process for the generation of these APIs.
We do have some worries, though, which we need to discuss before finalising on this route:
1) Will the additional admin overhead be worth the gain ?
The feeling on the call was that this approach has potential to get more people involved, and worst case, leaves us in the current situation with an active WG and an inactive RG.
2) Will this approach really help us to engage and attract new people ?
We are hoping that the ability to spawn small, tightly-focussed groups will help make it easier to attract people and focus them on the API development.
We are planning to have another conference call next Wednesday at 1400 GMT (same time as today's call), to continue discussion of this, and on Thursday wish to send the decision of the group to the GFSG so that they may discuss it before GGF.
I know this is a short time-frame, but we need to finalise this issue soon, and comments from the wider group are essential. Please speak up one way or another in the next few days as to whether we should go forward with this plan, or ask the GFSG to just 'flip-the-bit' as per the original discussions. Any other comments or suggestions would be great.
Even a response such as 'no, just flip the bit', or 'go for it' would be helpful.
We have assurance from the ADs that whichever way we decide to go, the transition will be quick and relatively painless.
Cheers,
Tom
-- +-----------------------------------------------------------------+ | Andre Merzky | phon: +31 - 20 - 598 - 7759 | | Vrije Universiteit Amsterdam (VU) | fax : +31 - 20 - 598 - 7653 | | Dept. of Computer Science | mail: merzky@cs.vu.nl | | De Boelelaan 1083a | www: http://www.merzky.net | | 1081 HV Amsterdam, Netherlands | | +-----------------------------------------------------------------+
On Sat, 4 Feb 2006, Andre Merzky wrote:
I think the name SAGA-CORE-WG is misleading, as the strawman does not really define a Core API. That role, defining the core, look and feel, and basic paradigms, would rather fit the target of the RG.
Definitely. In the future the SAGA WGs will presumably be named according to the functional area they are producing an API for, e.g. SAGA-GridCPR-WG, but that would be unwieldy for this initial group, so I had to suggest something 8-)
However, during yesterdays discussion we could not come up with a better name, really. Just so you all have something to laugh at, I'll repeat some suggestions below. Well, the point of this mail is: if you can think of a suitable group name, please speak up!
iSAGA (very fashionable, for initial SAGA)
Perhaps SAGA-Init-WG ? Or SAGA-Kickstart-WG, or SAGA-FLJST-WG (Files Logical-files Jobs Streams Tasks) ?
SAGA-PRIMER (obvious)
Sounds more like a group producing a programming primer.
SSAGA (confusing, but stands for Super simple API...)
Too generic; it suggests it is doing something super within SAGA, rather than 'just' being a group producing a set of APIs for a well defined set of funtional areas.
My personal favourite for today is (cling on tight)
SAGA-WG
Well, that might be confusing with SAGA-RG, but what the heck - it describes what we do best I think...
I think it would be confusing, not so much with the RG, but more in that the name suggests it is producing the whole SAGA API, and not just a well defined set of APIs. Cheers, Tom
Quoting [Tom Goodale] (Feb 04 2006):
iSAGA (very fashionable, for initial SAGA)
Perhaps SAGA-Init-WG ?
Yes, perhaps - not bad :-)
Or SAGA-Kickstart-WG, or SAGA-FLJST-WG (Files Logical-files Jobs Streams Tasks) ?
ARGHL! :-P If you can't speak it three times, in 5 seconds, its out I'd say... ;-) Agree with below. Cheers, Andre.
SAGA-PRIMER (obvious)
Sounds more like a group producing a programming primer.
SSAGA (confusing, but stands for Super simple API...)
Too generic; it suggests it is doing something super within SAGA, rather than 'just' being a group producing a set of APIs for a well defined set of funtional areas.
My personal favourite for today is (cling on tight)
SAGA-WG
Well, that might be confusing with SAGA-RG, but what the heck - it describes what we do best I think...
I think it would be confusing, not so much with the RG, but more in that the name suggests it is producing the whole SAGA API, and not just a well defined set of APIs.
Cheers,
Tom
-- +-----------------------------------------------------------------+ | Andre Merzky | phon: +31 - 20 - 598 - 7759 | | Vrije Universiteit Amsterdam (VU) | fax : +31 - 20 - 598 - 7653 | | Dept. of Computer Science | mail: merzky@cs.vu.nl | | De Boelelaan 1083a | www: http://www.merzky.net | | 1081 HV Amsterdam, Netherlands | | +-----------------------------------------------------------------+
On Sat, Feb 04, 2006 at 10:30:05AM +0100, Andre Merzky wrote:
I think the name SAGA-CORE-WG is misleading, as the strawman does not really define a Core API. That role, defining the core, look and feel, and basic paradigms, would rather fit the target of the RG.
I'd like to disagree, violently! IF the current API scope of the strawman is supposed to make any sense, then as the CORE functionality, the absolute required minimum needed. (files and job spawning, and the stuff needed to support this -- very roughly speaking) If it is not this, then I would not see the justification for the strawman's contents. Thilo PS: Andre, this is "core" as in "the very essential", not as in "rock solid, made for eternity". -- Thilo Kielmann http://www.cs.vu.nl/~kielmann/
Ah, I knew it wouldn't be _that_ easy ;-) Well, what is the CORE of Grid capabilities is in the eye of the beholder I think. Certainly, the strawman covers the core capabilities from the SAGA use cases. However, if you ask some OGSA guy about the core of Grid capabilities, I am fairly sure that stream multiplexing will be very much absent on his list... So, for us it certainly is the core of things we need - in the grandbright future of SAGA scope it is probably not. But, actually, whatever - its just naming after all... ;-) As long as we know what we want to achieve it does not matter. (But I would hate to hear "you need to do service management, that _definitely_ belongs to the core of Grids"). Cheers, Andre. Quoting [Thilo Kielmann] (Feb 04 2006):
On Sat, Feb 04, 2006 at 10:30:05AM +0100, Andre Merzky wrote:
I think the name SAGA-CORE-WG is misleading, as the strawman does not really define a Core API. That role, defining the core, look and feel, and basic paradigms, would rather fit the target of the RG.
I'd like to disagree, violently!
IF the current API scope of the strawman is supposed to make any sense, then as the CORE functionality, the absolute required minimum needed. (files and job spawning, and the stuff needed to support this -- very roughly speaking)
If it is not this, then I would not see the justification for the strawman's contents.
Thilo
PS: Andre, this is "core" as in "the very essential", not as in "rock solid, made for eternity".
-- +-----------------------------------------------------------------+ | Andre Merzky | phon: +31 - 20 - 598 - 7759 | | Vrije Universiteit Amsterdam (VU) | fax : +31 - 20 - 598 - 7653 | | Dept. of Computer Science | mail: merzky@cs.vu.nl | | De Boelelaan 1083a | www: http://www.merzky.net | | 1081 HV Amsterdam, Netherlands | | +-----------------------------------------------------------------+
I think we have an advantage: we have the applications' view as stated in our use cases. E.g., service management might be what (OGSA) services providers care about, our applications however, don't. Remember the "A" at the end of "SAGA"? Thilo On Sat, Feb 04, 2006 at 04:14:44PM +0100, Andre Merzky wrote:
X-Original-To: kielmann@localhost Delivered-To: kielmann@localhost.cs.vu.nl Delivered-To: grdfm-saga-rg-outgoing@mailbouncer.mcs.anl.gov X-Original-To: grdfm-saga-rg@mailbouncer.mcs.anl.gov Delivered-To: grdfm-saga-rg@mailbouncer.mcs.anl.gov Date: Sat, 4 Feb 2006 16:14:44 +0100 From: Andre Merzky
To: Thilo Kielmann Cc: saga-rg@ggf.org Subject: Re: [saga-rg] Research Group / Working Groups X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at mailbouncer.mcs.anl.gov X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at mailbouncer.mcs.anl.gov Ah, I knew it wouldn't be _that_ easy ;-)
Well, what is the CORE of Grid capabilities is in the eye of the beholder I think. Certainly, the strawman covers the core capabilities from the SAGA use cases.
However, if you ask some OGSA guy about the core of Grid capabilities, I am fairly sure that stream multiplexing will be very much absent on his list...
So, for us it certainly is the core of things we need - in the grandbright future of SAGA scope it is probably not.
But, actually, whatever - its just naming after all... ;-) As long as we know what we want to achieve it does not matter. (But I would hate to hear "you need to do service management, that _definitely_ belongs to the core of Grids").
Cheers, Andre.
Quoting [Thilo Kielmann] (Feb 04 2006):
On Sat, Feb 04, 2006 at 10:30:05AM +0100, Andre Merzky wrote:
I think the name SAGA-CORE-WG is misleading, as the strawman does not really define a Core API. That role, defining the core, look and feel, and basic paradigms, would rather fit the target of the RG.
I'd like to disagree, violently!
IF the current API scope of the strawman is supposed to make any sense, then as the CORE functionality, the absolute required minimum needed. (files and job spawning, and the stuff needed to support this -- very roughly speaking)
If it is not this, then I would not see the justification for the strawman's contents.
Thilo
PS: Andre, this is "core" as in "the very essential", not as in "rock solid, made for eternity".
-- +-----------------------------------------------------------------+ | Andre Merzky | phon: +31 - 20 - 598 - 7759 | | Vrije Universiteit Amsterdam (VU) | fax : +31 - 20 - 598 - 7653 | | Dept. of Computer Science | mail: merzky@cs.vu.nl | | De Boelelaan 1083a | www: http://www.merzky.net | | 1081 HV Amsterdam, Netherlands | | +-----------------------------------------------------------------+
-- Thilo Kielmann http://www.cs.vu.nl/~kielmann/
Quoting [Thilo Kielmann] (Feb 04 2006):
I think we have an advantage: we have the applications' view as stated in our use cases.
E.g., service management might be what (OGSA) services providers care about, our applications however, don't.
Remember the "A" at the end of "SAGA"?
Ah, right, there's an 'A' ;-) Ok, fair point. But still I think that our idea of a core might not be long lasting. Whatever :-) Cheers, Andre.
Thilo
On Sat, Feb 04, 2006 at 04:14:44PM +0100, Andre Merzky wrote:
Ah, I knew it wouldn't be _that_ easy ;-)
Well, what is the CORE of Grid capabilities is in the eye of the beholder I think. Certainly, the strawman covers the core capabilities from the SAGA use cases.
However, if you ask some OGSA guy about the core of Grid capabilities, I am fairly sure that stream multiplexing will be very much absent on his list...
So, for us it certainly is the core of things we need - in the grandbright future of SAGA scope it is probably not.
But, actually, whatever - its just naming after all... ;-) As long as we know what we want to achieve it does not matter. (But I would hate to hear "you need to do service management, that _definitely_ belongs to the core of Grids").
Cheers, Andre.
Quoting [Thilo Kielmann] (Feb 04 2006):
On Sat, Feb 04, 2006 at 10:30:05AM +0100, Andre Merzky wrote:
I think the name SAGA-CORE-WG is misleading, as the strawman does not really define a Core API. That role, defining the core, look and feel, and basic paradigms, would rather fit the target of the RG.
I'd like to disagree, violently!
IF the current API scope of the strawman is supposed to make any sense, then as the CORE functionality, the absolute required minimum needed. (files and job spawning, and the stuff needed to support this -- very roughly speaking)
If it is not this, then I would not see the justification for the strawman's contents.
Thilo
PS: Andre, this is "core" as in "the very essential", not as in "rock solid, made for eternity".
-- "So much time, so little to do..." -- Garfield
participants (3)
-
Andre Merzky
-
Thilo Kielmann
-
Tom Goodale