As far as I know in old versions of Boost boost::mutex implementation for Windows was done using critical sections. But in the newest version of Boost 1.51 I di
Isn't it wonderful that by using boost we always have best approach with no change?
In new version of boost, boost::mutex is implemented as an spinlock but with the help of a windows event to avoid busy wait and that event will only created if needed, thus it is very light weight and have a very high performance and also enable boost to use this light weight mutex for timed wait! I think this is excellent