> I'd love to see a reproducible deadlock test, even if it had to run > for awhile. I may have misunderstood you, but the PROVOKE-DEADLOCK test does just this and leads to deadlocks quickly. Leslie