Description
The current history-following algorithm determines the paths that a line of history would be *expected to be* found in a given revision, but doesn't treat copies as gaps in the historical sequence. If some other object fills those gaps at the same expected path, the client will find the wrong object. My next comment will contain a recipe script.