3 Commits (2322bb7d86b9d288f2558e3d22af5fd60c0f38fd)

Author	SHA1	Message	Date
hanna	2322bb7d86	Workaround: use a single ReferenceIterator for an entire micromanaged traversal. We'll have to ... do something about ReferenceIterator thread safety later. git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@361 348d0f76-0448-11de-a6fe-93d51630548a	2009-04-10 20:50:28 +00:00
hanna	8a1207e4db	Bringing up scaffolding for integration of locus traversals by reference with Aaron's data source code. ... Reverts to original TraverseByLociByReference behavior unless a special combination of command-line flags are used. Lightly tested at best, and major flaws include: - MicroManager is not doing MicroScheduling right now; it's driving the traversals. - New database-ish data providers imply by their interface that they're stateless, but they're highly stateful. - Using static objects to circumvent encapsulation. - Code duplication is rampant. - Plus more! git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@346 348d0f76-0448-11de-a6fe-93d51630548a	2009-04-09 20:28:17 +00:00
depristo	cfee59e0e6	New type hierarchy for Traversals. There's a new package to hold them (traversals) and an easy system to create new ones. We are now one step closer to supporting the execution manager (a totally non-functional version is included here) that actually executes walkers in parallel using N threads. ... git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@214 348d0f76-0448-11de-a6fe-93d51630548a	2009-03-27 15:40:45 +00:00