Re: [Boost-users] Boost interprocess named mutexes

Mike Spertus escribió:

Sorry, I misread that doc page. You are of course correct. What do you think of my suggestion to use a Windows mutex to back a boost interprocess named mutex? Since "abandoned Windows mutexes" are automatically released by Windows, we see substantially fewer hangs when we use Windows mutexes.

Windows named mutexes have reference-counted semantics, so they disappear when the last attached process dies, so they are not ompatible with POSIX lifetime semantics. Best, Ion