I am seeing the following problems with Boost interprocess named mutexes on Windows. 1. As mentioned in my other threads, interprocess mutexes do not provide mutual exclusion between processes run by different users. The reason for this is the shared memory mappings backing the mutexes are placed in different directories. 2. If a process crashes while holding a mutex, no other process can subsequently acquire the mutex. 3. Although the documentation says that mutexes have process persistence, they appear to have filesystem persistence. This means that is a process or system crashes while the mutex is held, the mutex is permanently unavailable (without deleting it). Are these accurate statements? If so, could these be addressed by having Boost::interprocess use Win32 named mutexes (which I understand to have none of these defects) like it uses pthread mutexes on Linux? Thanks, Mike