Hi, to disable the excuse 'too long' for not reading the strawman API spec, we created a short version of the spec. It does not contain use case relation, notes, examples, and the detailed deiscussion of the API objects and methods, but only - intro etc - rationale for each subsystem - SIDL for each subsystem You can find the short version as pdf and txt on the WIKI: http://wiki.cct.lsu.edu/saga/ check the last links for the v0.2 version of the doc on the front page. However, please check the complete doc on any questions before complaining on the list, in particular when in doubt about call semantics. Cheers, Andre. -- +-----------------------------------------------------------------+ | 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 | | +-----------------------------------------------------------------+
to disable the excuse 'too long' for not reading the strawman API spec, we created a short version of the spec.
Having recently read through a recent 'compact' version of the specification I think there is a serious need to focus. Although there is a lot of good stuff in there it is not IMHO 'simple'. Steven -- ---------------------------------------------------------------- Dr Steven Newhouse Tel:+44 (0)2380 598789 Deputy Director, Open Middleware Infrastructure Institute (OMII) Suite 6005, Faraday Building (B21), Highfield Campus, Southampton University, Highfield, Southampton, SO17 1BJ, UK
Is it merily confusing (== not simple) because its so much, or are there parts which are not simple in itself? A. Quoting [Steven Newhouse] (Aug 01 2005):
to disable the excuse 'too long' for not reading the strawman API spec, we created a short version of the spec.
Having recently read through a recent 'compact' version of the specification I think there is a serious need to focus. Although there is a lot of good stuff in there it is not IMHO 'simple'.
Steven
-- +-----------------------------------------------------------------+ | 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 | | +-----------------------------------------------------------------+
Hmm, I think we should discuss this at tomorrows phone call... Cheers, Andre. Quoting [Andre Merzky] (Aug 01 2005):
Is it merily confusing (== not simple) because its so much, or are there parts which are not simple in itself?
A.
Quoting [Steven Newhouse] (Aug 01 2005):
to disable the excuse 'too long' for not reading the strawman API spec, we created a short version of the spec.
Having recently read through a recent 'compact' version of the specification I think there is a serious need to focus. Although there is a lot of good stuff in there it is not IMHO 'simple'.
Steven
-- +-----------------------------------------------------------------+ | 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 | | +-----------------------------------------------------------------+
Is it merily confusing (== not simple) because its so much,
The basic entry point involves too many 'simple' interfaces. If I just want to submit a job, what do I want to deal with? The session (probably not) & security (probably yes) APIs? Steven -- ---------------------------------------------------------------- Dr Steven Newhouse Tel:+44 (0)2380 598789 Deputy Director, Open Middleware Infrastructure Institute (OMII) Suite 6005, Faraday Building (B21), Highfield Campus, Southampton University, Highfield, Southampton, SO17 1BJ, UK
Well, the question is what do you want to do, really. Just submit a job, just copy a file, just replicate one, just open a stream... In itself the use cases are simple, if you take them all together its getting a large API. Session handle and security are designed to have sensible default values - so you don't need to touch them. We should make that very clear. E.g., if there is a X509 proxy for your user id lying around, the default session handle (which you won't even see) should have a X509 context attached automatically. Now, the qeustion really is: is it still too complex, or are we cleverly hiding the simplicity? :-D I assume that I am somewhat blinkered due to my daily dose of SAGA - I think its blindingly simple, and we just need good examples (e.g. map API to our use cases) to make that obvious. I might be wrong though...? Andre. Quoting [Steven Newhouse] (Aug 03 2005):
Date: Wed, 03 Aug 2005 16:40:35 +0100 From: Steven Newhouse <sjn5@doc.ic.ac.uk> To: Andre Merzky <andre@merzky.net> CC: Simple API for Grid Applications WG <saga-rg@ggf.org> Subject: Re: [saga-rg] Short Strawman versions
Is it merily confusing (== not simple) because its so much,
The basic entry point involves too many 'simple' interfaces. If I just want to submit a job, what do I want to deal with? The session (probably not) & security (probably yes) APIs?
Steven -- +-----------------------------------------------------------------+ | 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 | | +-----------------------------------------------------------------+
My guess would be that the whole SAGA effort requires a careful balance between simplicity and usefulness. With this respect, I believe that the SAGA group is on track: starting from the use cases, working towards getting exactly these use cases simple. Deciding about where to draw the border of things that should be kept out of SAGA for the sake of simplicity is the crux of the balance. Comparing to the MPI standard, I believe there are 2 lessons to be learned: (at least ;-) 1. make the API modular (SAGA does that already) Many people write (their first) MPI codes just with MPI_Init and Finalize, and with MPI_Send and Receive. (and get only later to the more flexible things that require further understanding.) To this respect, the very essential subset of SAGA should be identified and supported by implementations that "do the right thing automagically" by having good default values (e.g. for security contexts) 2. there should be a "SAGA cookbook" MPI failed on this respect and good books came out only late, delaying the uptake of MPI significantly. I would thus suggest that a SAGA cookbook gets written as soon as possible, maybe as part of the Wiki? My 2 cents, Thilo On Thu, Aug 04, 2005 at 08:57:16AM +0200, Andre Merzky wrote:
Well, the question is what do you want to do, really. Just submit a job, just copy a file, just replicate one, just open a stream... In itself the use cases are simple, if you take them all together its getting a large API.
Session handle and security are designed to have sensible default values - so you don't need to touch them. We should make that very clear. E.g., if there is a X509 proxy for your user id lying around, the default session handle (which you won't even see) should have a X509 context attached automatically.
Now, the qeustion really is: is it still too complex, or are we cleverly hiding the simplicity? :-D
I assume that I am somewhat blinkered due to my daily dose of SAGA - I think its blindingly simple, and we just need good examples (e.g. map API to our use cases) to make that obvious. I might be wrong though...?
Andre.
Quoting [Steven Newhouse] (Aug 03 2005):
Date: Wed, 03 Aug 2005 16:40:35 +0100 From: Steven Newhouse <sjn5@doc.ic.ac.uk> To: Andre Merzky <andre@merzky.net> CC: Simple API for Grid Applications WG <saga-rg@ggf.org> Subject: Re: [saga-rg] Short Strawman versions
Is it merily confusing (== not simple) because its so much,
The basic entry point involves too many 'simple' interfaces. If I just want to submit a job, what do I want to deal with? The session (probably not) & security (probably yes) APIs?
Steven -- +-----------------------------------------------------------------+ | 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] (Aug 04 2005):
2. there should be a "SAGA cookbook" MPI failed on this respect and good books came out only late, delaying the uptake of MPI significantly.
I would thus suggest that a SAGA cookbook gets written as soon as possible, maybe as part of the Wiki?
I think we should do that as soon as the API and the first implementation stabelizes somewhat - otherwise the maintainance effort is too high. But you are right, we should have that ready with the first releases... A.
My 2 cents,
Thilo
On Thu, Aug 04, 2005 at 08:57:16AM +0200, Andre Merzky wrote:
Well, the question is what do you want to do, really. Just submit a job, just copy a file, just replicate one, just open a stream... In itself the use cases are simple, if you take them all together its getting a large API.
Session handle and security are designed to have sensible default values - so you don't need to touch them. We should make that very clear. E.g., if there is a X509 proxy for your user id lying around, the default session handle (which you won't even see) should have a X509 context attached automatically.
Now, the qeustion really is: is it still too complex, or are we cleverly hiding the simplicity? :-D
I assume that I am somewhat blinkered due to my daily dose of SAGA - I think its blindingly simple, and we just need good examples (e.g. map API to our use cases) to make that obvious. I might be wrong though...?
Andre.
Quoting [Steven Newhouse] (Aug 03 2005):
Date: Wed, 03 Aug 2005 16:40:35 +0100 From: Steven Newhouse <sjn5@doc.ic.ac.uk> To: Andre Merzky <andre@merzky.net> CC: Simple API for Grid Applications WG <saga-rg@ggf.org> Subject: Re: [saga-rg] Short Strawman versions
Is it merily confusing (== not simple) because its so much,
The basic entry point involves too many 'simple' interfaces. If I just want to submit a job, what do I want to deal with? The session (probably not) & security (probably yes) APIs?
Steven -- +-----------------------------------------------------------------+ | 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 | | +-----------------------------------------------------------------+
-- +-----------------------------------------------------------------+ | 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'd like to second Thilo's comments. I've often used the "six calls of MPI" analogy for what SAGA should be defining for grid apps. The current SAGA API doc, however, really is a reference manual that defines the API in (mostly) complete (i.e., "gory") detail. That said, what we need is a "SAGA for Beginners" document or "cookbook" that has lots of example programs -- starting with very, very simple programs and progressing to programs add more functionality. --Craig At 02:37 AM 8/4/2005, Thilo Kielmann wrote:
My guess would be that the whole SAGA effort requires a careful balance between simplicity and usefulness. ...
Comparing to the MPI standard, I believe there are 2 lessons to be learned: (at least ;-)
1. make the API modular (SAGA does that already) Many people write (their first) MPI codes just with MPI_Init and Finalize, and with MPI_Send and Receive. (and get only later to the more flexible things that require further understanding.)
To this respect, the very essential subset of SAGA should be identified and supported by implementations that "do the right thing automagically" by having good default values (e.g. for security contexts)
2. there should be a "SAGA cookbook" MPI failed on this respect and good books came out only late, delaying the uptake of MPI significantly.
I would thus suggest that a SAGA cookbook gets written as soon as possible, maybe as part of the Wiki?
My 2 cents,
Thilo
On Thu, Aug 04, 2005 at 08:57:16AM +0200, Andre Merzky wrote:
Well, the question is what do you want to do, really. Just submit a job, just copy a file, just replicate one, just open a stream... In itself the use cases are simple, if you take them all together its getting a large API.
Session handle and security are designed to have sensible default values - so you don't need to touch them. We should make that very clear. E.g., if there is a X509 proxy for your user id lying around, the default session handle (which you won't even see) should have a X509 context attached automatically.
Now, the qeustion really is: is it still too complex, or are we cleverly hiding the simplicity? :-D
I assume that I am somewhat blinkered due to my daily dose of SAGA - I think its blindingly simple, and we just need good examples (e.g. map API to our use cases) to make that obvious. I might be wrong though...?
Andre.
Quoting [Steven Newhouse] (Aug 03 2005):
Date: Wed, 03 Aug 2005 16:40:35 +0100 From: Steven Newhouse <sjn5@doc.ic.ac.uk> To: Andre Merzky <andre@merzky.net> CC: Simple API for Grid Applications WG <saga-rg@ggf.org> Subject: Re: [saga-rg] Short Strawman versions
Is it merily confusing (== not simple) because its so much,
The basic entry point involves too many 'simple' interfaces. If I just want to submit a job, what do I want to deal with? The session (probably not) & security (probably yes) APIs?
Steven -- +-----------------------------------------------------------------+ | 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/
participants (4)
-
Andre Merzky -
Craig Lee -
Steven Newhouse -
Thilo Kielmann