[kepler-dev] [Bug 3558] New: - Store each workflow run trace in it's own directory

bugzilla-daemon at ecoinformatics.org bugzilla-daemon at ecoinformatics.org
Thu Oct 23 16:33:32 PDT 2008


           Summary: Store each workflow run trace in it's own directory
           Product: Kepler
           Version: 1.0.0
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: general
        AssignedTo: mcphillips at ecoinformatics.org
        ReportedBy: mcphillips at ecoinformatics.org
         QAContact: kepler-dev at ecoinformatics.org

The ppod-gui module includes enhancements to the Kepler GUI for browsing and
viewing the traces of previously run workflows.  Each trace is a single,
XML-formatted file, and the traces are organized in folders named according to
the workflows run.  These folders corresponding to actual directories on the
user's machine.  Double-clicking on a trace opens it in the provenance browser.

I suggest that instead of putting all the traces of runs of a particular
workflow in the same directory, we create a new directory for each run (in the
directory named for the workflow) and place the trace there.  This would
support: (a) including multiple traces output from a single run (possible with
COMAD); (b) storing a copy of the workflow in the run directory so that the
specification of the executed workflow is not lost; (c) using this directory as
the default location for other files produced during the run (and possibly
temporary directories holding intermediate files useful for resuming an aborted
run, say); (d) summary reports generated for the run; (e) copies of input data;

Adding another level of directories would add to the work of navigating to the
latest trace, but this would not be a problem if we automatically opened the
trace using the provenance browser at the end of each run (as suggested in Bug
3546), and if we provided an additional view of recent traces that hid this
extra nesting.

More information about the Kepler-dev mailing list