|
Aug 13, 2008
Participants: Altintas, Barseghian, Crawl, Jones, Ludäscher, Potter, Schildhauer, Schultz, Wang
- barseghian
- added a provenance class to just write tokens to disk (an attempt at 'intermediate data products'), usefulness debatable, probably won't pursue
- 'archive end-to-end run' usecase - looking at old smart rerun code as part of the alternative path, trouble getting it to work
- conclusion: focus on getting 'archive a complete end to end run' done, delay smart rerun for now
- hoping to move Baskett Slough station to HJ Andrews, discussing with Andrews people
- todo: email out request for additions to EIM paper, and where these might go
- crawl
- ocean usecase, now in one workflow.
- todo: get peter's input on where we're going with this, and its usefulness
- general kepler discussion of workflow hierarchical depth and difficulty with keeping track of where one is at. Should, in general, depth be limited to 2 levels?
- working on EIM sst poster with Ilkay.
- wang
- Current use cases:
- Large Parameter Sweeps and Whole workflow detached execution and monitoring with HosseiniSimulationWorkflow proposed by Parviez Hosseini
- C Code Porting
- same workflow, multiple execution with different parameters, remote execution in cluster.
- more use case needed.
- todo:
- post paper of plans for distributed framework
- Identify and Fix a typical use case for implementation.
- Detailed design and implementation.
|
