Meeting Notes March 08

This is version 8. It is not the current version, and thus it cannot be edited.
[Back to current version]   [Restore this version]

Milestones and Tasks

Terrestrial wf

1. PAR wf, reorganize par-data-clean step, replace with generic actors (Elementwise: min, max, is.na; UnivariateDistribution: min, max, range, interquartile range, mean, variance, median, stddev, historgram, densityplot, autocorrelation, partial autocorrelation; bivariate: correlation; statistical model fit: lm, glm, gam; filter actors) (Derik)
2. Parviez' workflow model for wrapping a simulation using CommandLineActor -- review annotations and resolve issues
3. Terr wf -- need to select data based on spatial and temporal bounds
4. Parameter sweep example workflow (define reqs of param sweep actor, implement actor after checking existing actors and directors)
5. Review user manual section on Command Line Exec actor (section 5.5) (Parviez)

SST wf

1. Layman version (without technical and scientific buzwords) of the description of the goals for each step and workflow(couple of paragraphs) (April 4th; 1 week) (Ilkay + Mark)
2. Peter will write up his realization of why this is useful on top of just matlab with a few bullets(April 4th; 1 week) (provenance, chaining what is available, ...) (Peter)
3. Conceptual workflow with input/output details and placeholders for different computational components (April 9th; 2 weeks) (Ilkay and Dan)
4. Design the schema ad build a relational database for the match up datasets (April 11th; 2 weeks) (Dan and Ilkay)
5. Recollection of requirements and comments (April 17th; 3 weeks) (Ilkay and everyone on SST)
6. Working version0 (May 15th; 1.5 months - 4 weeks for implementation) (Ilkay and Dan)
7. Definition of additional analysis steps (June 1st; 2 months) (Ilkay with input from Peter)
8. GHRSST Meeting (June 9-13, 2008; France) presentation and demonstration (June 9th; 2 months) (Peter)

Development tasks

1. Develop user interface and data structures for choosing a coverage and subsetting data using that coverage specification; may be a sampling actor and/or search (when: April 30 who: )
2. Document OPeNDAP actor, add to tree, remove ODC feature, assign icon, test for release (Nathan (with Kirsten)), (Next week))
3. Matlab actor -- prepare for release, get library loading working, remove MatlabExpression version of actor (cretae bug to merge this as a fallback to JNI), make actor use TextArea for script, check documentation, complete test workflows (Daniel, with help from Kirsten, by mid April)
4. Change ProvenanceRecorder to use local hsql db, turned on by default?, possibly output to text format on demand (Daniel, by mid April for Kepler release)
5. Combine EML and DAP DataSource actors
6. Create data location actor (finds data and spits out URL)
7. Pass data by-reference in addition to by-value
8. DataTurbine actor: PortParam bug, change name, gap filling feature (probably separate actor if doing interpolation), only output data on connected ports, assign icon
9. CSV file reader (check name and documentation (tab, comma, csv should all return actor), ignore headers and footers (check the existing R Read.Table actor)
10. Add DateTime token type
11. Add DataFrame token type
12. Modify Command Line Exec actor to trap errors so that workflow can continue after error or exception or core dump (Daniel)
13. Remove External Execution actor from actor list, docs, and maybe code base, and/or maybe consolidate COmmandLineExec with External Execution (Daniel)