Maybe the mutexes don't guarantee it on all platforms? I found this when looking for stuff about mutexes on linux: http://lxr.linux.no/source/Documentation/robust-futexes.txt And it appears to guarantee robustness. Cheers Lars Ion Gaztañaga wrote:
Lars Hagström wrote:
How about adding (as an option?) this kind of behaviour to the emulated mutexes in interprocess?
I could provide a macro to use emulated mutexes/conditions instead of native ones.
The Posix mutexes I assume guarantee some kind of robustness, so that if a process crashes its locks are released. by adding the below behaviour we get reasonably close to that behaviour for the emulated mutexes.
I don't Posix mutexes provide much robustness but they provide much higher performance, since emulation is based on atomic compare and exchange plus spin-waits.
Anyway, I definitely need some help on Mac OS 10.5. I don't know if anyway has used those process-shared mutexes in practice and if they work at all.
Cheers Lars
Regards,
Ion _______________________________________________ Boost-users mailing list Boost-users@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users