[SAGA-RG] notes from the Java binding meeting at OGF22

[Thilo Kielmann]

Andre's notes pretty much speak for themselves ;-)

Thilo

----- Forwarded message from Andre Merzky <andre at merzky.net> -----

> From: Andre Merzky <andre at merzky.net>
> To: Thilo Kielmann <kielmann at 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/