2010/4/27 Ingo Weinhold <ingo_weinhold@xxxxxx>: > Not sure, if this is an actual problem, but there's a race condition: If the > rescan thread leaves the loop and is preempted before resetting fThreadId, a > concurrent Rescan() would not start a new thread. IOW the rescan request > would not be processed in such a situation. The thread start and exit parts are now protected by a lock in r36588. I hope it works well now. Bye, Jérôme