Bug #5790
closedNot all remote run-kars appear in Workflow Run Manager
0%
Description
I believe the Workflow Run Manager is not displaying all the remote run-kars that it should. E.g. "04-HelloWorld-execId6_20120830-204114819PDT,kar" (as seen in Component Search). My guess would be something's going wrong in the rather hideous ProvenanceStore.downloadAllWorkflowRuns method I wrote. This method avoids the very slow process of downloading all run-kars to get at their contained WorkflowRun files, and instead reassembles, in a very fragile way, WorkflowRuns from fragments returned by a query on the KarXMLs in metacat. At the time I didn't know of a way to query for/extract from results the entire WorkflowRun xmls in a cleanly delimited way, but iirc later Matt and/or Ben told me there is. Look into this...
Updated by ben leinfelder over 11 years ago
A way to extract full fragments of XML in a query result!? I would love to know how.
Updated by ben leinfelder over 11 years ago
Talked with Derik about this today and suggested issuing multiple queries, each targeting a specific part of the runkar moml @name/@value pairs using a predicate xpath condition in the query. You'd then aggregate these multiple search results to build the runkar preview. I think this is the best way to guarantee query consistency since we can're rely on the @name/@value pairs being in any particular order let alone corresponding to one another.
Updated by Derik Barseghian over 11 years ago
After some testing and further discussion w/ Jing and Ben, it turns out the condition section of the query is just for determining which docids to return, with returnfields being returned from those, so predicates there won't help my usecase. However you can use predicates in returnfields (something we didn't think we supported), which is great. I'm in the process of refactoring my query to do this.