Hi all, After confirming that the subsystem component wasn't responsible (sorry John!), John Ross has suggested that the second set of changes under ARIES-858, introduced in build #1470, might be responsible for the test failures we're seeing. Since those changes affected the concurrency in NamespaceHandlerRegistryImpl, and the test failures seem to be an intermittent failure to register some namespace handlers, this seems like a good guess to me.
The bad news is that if these changes do cause concurrency issues in blueprint, the 0.3.2 release will be have the defect too. On the bright side, at least it's only an intermittent failure! Guillaume, what problem were these concurrency changes trying to fix? I'm wondering if we're better off just going back to the NamespaceHandlerRegistryImpl of a month ago, since that seemed to be more reliable in our tests. Holly
