February 2008 - saga-rg - lists.ogf.org

EGEE glite and SAGA
by Ashiq Anjum 07 Mar '08

07 Mar '08

Hi, We got a new project in which we have to create a generic services layer which should make client applications independent from a particular Grid middleware. I want to use SAGA API to access glite services. Does any one has a glite adapter for SAGA? Suggestions and experiences are welcomed to achieve this goal. cheers, Ashiq Anjum

5 11

Java binding: object factories
by Thilo Kielmann 27 Feb '08

27 Feb '08

All, from this week's meeting on the Java language binding, there is one issue left over that (to me) sounds like a matter of taste... For all object factories, we have two flavors, one with an explicit Session parameter, and one with an implicit (default) Session. Example: with the implicit default Session: File file = FileFactory.createFile(....); for the explicit Session parameters, there are two proposals: (a) have the create* methods in the factory classes, resulting in two create* methods each, with slightly different signature (Session parameter or not) example: File file = FileFactory.createFile(session, ....); (b) make the Session class also a factory and add all the create* methods there. example: File file = session.createFile(....); None of the people I've asked so far seems to have a strong opinion on this. Personally, I have a slight bias towards (a) as it seems to be conceptionally simpler. (objects are always created by the factories...) Also, that's as it is in the current language binding. I hereby suggest that we keep it "as is" unless somebody can convince us of the superiority of option (b). (Of course, we will change once convinced ;-) Regards, Thilo -- Thilo Kielmann http://www.cs.vu.nl/~kielmann/

2 1

notes from the Java binding meeting at OGF22
by Thilo Kielmann 27 Feb '08

27 Feb '08

Andre's notes pretty much speak for themselves ;-) Thilo ----- Forwarded message from Andre Merzky <andre(a)merzky.net> ----- > From: Andre Merzky <andre(a)merzky.net> > To: Thilo Kielmann <kielmann(a)cs.vu.nl> > Subject: notes > > :-) > > > OGF.22 SAGA-SESSION I: Java Language Bindings > > When: 9:30 - 11:00 > What: > > - IPR > - sessions during this OGF > - agenda accepted > > - SAGA spec landscape > > - design choices > > Q: Malcolm: on what level should concurrency be? Every call? > A: yes. But not enforcing execution order > > Q: why IOException, not SAGA_IO_Exception? > A: lets check... - yep, should do > > Q: where are infos encoded in ERRNO? > A: inside the exception. Extra field? > > Q: no permissions? really? > A: yes, really. ;-) > > Note: Java-7 will have permissions. > > - interfaces & factories > > Q: Malcolm: with boostrap loader, you can't mix two > implementations in one VM > A: right. We discussed this, but cuold not find a sensible use > case... > A: Daniel: Actually, does not prevent that, as you can use > multiple class loaders > > Q: Hidemoto: Which ways to set that property? > A: usual Java ways... > > Q: Why not have session.createFile (URL) > 'Solves' the default session problem (missing) > looks simple > gives same signature for both versions > A: Hmm, why not... > > - SAGAObject > > Q: do you need getType > A: could use isInstanceOf > > - saga.error > > Q: SagaError? What is the supposed action? > Q: Daniel: do NOT call it an Error > A: Hmmm... Yep, could use NoSuccess > If not, discuss again on mailing list. > > Q: Hidemoto: add 'Exception' to names. > Q: Exception -> SAGAException > Q: move exceptions to org.ogf.saga? > A: possibly yes to all of them > > - tasks: > > Q: task container: possibly get_task_type (cookie)? > A: not possible > > - job > > Q: Daniel: Task<Object> -> Task<?extendsObject> > A: aha > > - buffer > > Q: can you have impl managed buffers? > A: yes > > Q: Thilo: should we use java.nio.ByteBuffer > A: Hmm, dunno. > > Q: buffer should have long for size? > A: yes > > - summary > > Q: open points? > A: address all 'binding issue' tags in the core spec > > Q: intro sections? > A: remove intros if semantics is the same > A: Daniel: keep it so user does not need to jump so often > > -> tough call, probably should keep it. > > > Q: Shantenu: how many people intent to use that spec? > A: not many.... > > > > > -- > "We've got too much time to waste to stand around here doing things." > - Tigger ----- End forwarded message ----- -- Thilo Kielmann http://www.cs.vu.nl/~kielmann/

3 3

Session Notes
by Andre Merzky 27 Feb '08

27 Feb '08

Session: SAGA + DAIS We did not take detailed notes, but here is a short summary, and the road forward: It seems that the time for approaching a DAIS package in SAGA is about right, although both groups are short of resources. Exactly what abstractions are to be exposed by the SAGA API package remains a topic of discussion though. In order to solve that question, the DAIS folx will consolidate their use cases, and will provide us with versions focusing on the high level abstractions. Timeline for these use cases is end of April, we intent then to draft an API package for OGF23 in June. Cheers, Andre. -- "We've got too much time to waste to stand around here doing things." - Tigger

1 0

Additional session, tomorrow 9am
by Andre Merzky 26 Feb '08

26 Feb '08

Hi Folx, we had 5 good sessions at OGF22, well visited, and good result (notes will be posted soonish). However, we did not find time do discuss a couple of things, so we'll have an informal SAGA session tomorrow 9am in Ballroom A, at the XtreemOS booth. Agenda: - prioritize work on SAGA extensions - discuss and finalize message bus extension So, if you are interested, come along! :-) Same if you have other open topics... Cheers, Andre. -- "We've got too much time to waste to stand around here doing things." - Tigger

2 1

Service Type
by Fisher, SM (Steve) 24 Feb '08

24 Feb '08

Hi, I was just making the agreed chnages to the SD spec when I relaiszed that you said to use saga::<package>::<class> with package and class according to the implementation language. This won't work easily as means altering the string from the information service or wherever it comes from according to the chosen language. I think that a service type must be independent of the chosen language and so taken from the spec. With this in mind I want to include an example which would be the job_service on p180 of GFD.90. Should this be saga::job::job_service. Given that the package is actually saga.job then perhaps the best name would be saga.job.job_service i.e. <package>.<class> where the package always starts with "saga." Steve

2 3

OGF-22 plans...
by Andre Merzky 20 Feb '08

20 Feb '08

Hi all, please note that we created a wiki page with SAGA related sessions and meeting material on http://forge.ogf.org/short/saga-rg/ogf22 Feel free to add SAGA relevant information there. We will also add the session notes in the wiki (and post them to the list of course). Hope to see many of you in Boston! :-) Cheers, Andre. -- "We've got too much time to waste to stand around here doing things." - Tigger

1 0

Re: [SAGA-RG] [DAIS-WG] DAIS / SAGA...
by Andre Merzky 19 Feb '08

19 Feb '08

Hi Mario, I Cc this to the SAGA list, as it may help to prepare the session. Sorry for the cross posting folx! Quoting [Mario Antonioletti] (Feb 19 2008): > > Hi Andre, > Apologies for not having replied before - I will try to attend > this session. Are you expecting anything? i.e. a presentation or anything > else? Slightly concerned as, you said, we've had interactions before but > have had difficulties following these up due to time constraints. Let me > know if you want anything for the session and once again my apologies for > not having replied afore. No worries, nothing required apart from active participation. Its more a brainstorming, and an attempt to align our activities. The goal would be an answer to the following questions: - is there real potential for a DAIS package in SAGA? - if yes, can we come up with a reasonable work force to attack it? - if yes, can we come up with a reasonable time line? The main open points in respect to question one are, from our side: - what part of DAIS should actually be exposed on SAGA API level? - are different DAIS renderings exposed on API level, or can the API be uniform over the renderings? - is an extension mechanism required/possible for the different activity types? - should custom (i.e. domain specific) activity types be supported? How? - Can the DAIS data flow mechanism merged/combined with the SAGA task model (whcih has more potential as a basis for workflow)? You see, there are many general questions before we can start with drafting an actual API... Best, Andre. > > Mario > > > On Fri, 1 Feb 2008, Andre Merzky wrote: > >> Hi DAIS folx, >> >> we have been discussing DAIS/SAGA relationship a couple of >> times, w/o any specific actions items. We (the SAGA group) >> are now at a point where it would make really sense to >> approach the topic in earnest. >> >> So we scheduled a session at OGF22 for that topic, which >> will tentatively be scheduled on Tuesday 3:30 pm in 'Crispus >> Attucks' (its not listed in the program yet). It would be >> great if some of the DAIS regulars could join us, and >> protect us from doing really stupid things etc ;-) >> >> We will try to circulate some material before OGF, but I'd >> rather not promise that... >> >> Best, Andre. >> >> -- >> "We've got too much time to waste to stand around here doing things." >> - Tigger >> -- >> dais-wg mailing list >> dais-wg(a)ogf.org >> http://www.ogf.org/mailman/listinfo/dais-wg >> > > +-----------------------------------------------------------------------+ > |Mario Antonioletti:EPCC,JCMB,The King's Buildings,Edinburgh EH9 3JZ. | > |Tel:0131 650 5141|mario@epcc.ed.ac.uk|http://www.epcc.ed.ac.uk/~mario/ | > +-----------------------------------------------------------------------+ -- "We've got too much time to waste to stand around here doing things." - Tigger

1 0

Java language binding draft for discussion at OGF22
by Thilo Kielmann 19 Feb '08

19 Feb '08

All, at OGF21 we had a first discussion of the Java language binding. Since, we have been working hard on this item. The result is attached. Next week in Boston, we will have a session dedicated to this language binding. It would be great if some of you could have a look at it before. I have also put the draft into gridforge: http://forge.ogf.org/sf/go/doc15044?nav=1 or, project saga-rg, documents, java language binding See you next week in Boston! Thilo -- Thilo Kielmann http://www.cs.vu.nl/~kielmann/

1 0

SAGA SD Final?? Version
by Fisher, SM (Steve) 15 Feb '08

15 Feb '08

Hi, We have made the following changes to the Service Discovery spec: #1 Section 2.1.1: now includes reference to GLUE #2 Should the default VOs be chosen from the session object? As we are not able to come up with an intuitive syntax that allows the user to be able to request specific VOs or to take one from the security context we prefer to leave this as it is. #3 When there is Authz failure, should list_services throw 'NotEnoughPermission' exception? We have added the AuthorizationFailed exception (Incidentally I don't like the distinction between PermissionDenied and AuthorizationFailed in the core specification. I see no adequate justification for it - Steve) #4 To mention compliance to upcoming GLUE stds. (GLUE 2?). Whether GLUE upgrades affect the document? if so, appropriate mentioning required. (comments: similar to SAGA-core spec depending on JSDL) A footnote has been added The examples have also been updated to correspond precisely to our implementation. Note that we do now have a complete C++ implementation with gLite adapaters. The main code will be checked in as soon as Paventhan has write access to the repositoy and the adapter code will go into the gLite repository. We would like to see the document go to the OGF editor now unless somebody sees a problem. The .dvi file is attached. Steve and Paventhan

2 5