[kepler-dev] [Bug 5434] New: run-kar result from headless kepler.sh execution doesn't contain report artifacts

[bugzilla-daemon at ecoinformatics.org]

http://bugzilla.ecoinformatics.org/show_bug.cgi?id=5434

           Summary: run-kar result from headless kepler.sh execution
                    doesn't contain report artifacts
           Product: Kepler
           Version: 2.1.0
          Platform: Other
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: reporting
        AssignedTo: barseghian at nceas.ucsb.edu
        ReportedBy: barseghian at nceas.ucsb.edu
         QAContact: kepler-dev at kepler-project.org
   Estimated Hours: 0.0


This thread:
http://lists.nceas.ucsb.edu/kepler/pipermail/kepler-users/2011-June/002541.html
 prompted me to look into headless usage, esp wrt passing in workflow
parameters. It's unclear to me if passing in parameters is supported by ant
tasks like ant run-workflow-no-gui-server. So I began focusing on usage like:
./kepler.sh -runwf -nogui -myparam 7
/Users/derik/KeplerData/workflows/MyWorkflows/testwf.kar

If this testwf.kar contains a reportlayout (ROML), the reporting artifacts do
not end up in the run-kar (or in provenance associated w/ the run) after
execution if you specify parameters on the command line. If you don't pass in
parameters, the run-kar contains everything it should.

I tracked this down to the WorkflowManager changeListener not hearing the
change events on the model when it's changed for the parameter passed in. This
is because the NamedObj this listener gets added to is a different instance
than the one used a little bit later for execution. This is because the model
can't be found in the ModelDirectory at the execution point, because it's not
put in until that moment.

A solution to this bug is to call _openModel with the url to the temporary
unzipped workflow.xml file immediately after it's created in
KeplerConfigurationApplication(String[] args). Then the same NamedObj is
utilized, and the WorkflowManager listener doesn't get lost.

I want to look at this a bit more before checking in the fix, e.g. to see if
there's a better spot(s?) at which to put the model into the directory...

-- 
Configure bugmail: http://bugzilla.ecoinformatics.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.