Back in August 2005 there was a discussion on this list of the pros and cons of the existance of the try/catch(...) in thread_proxy(), which seemed to me to end with a pretty clear consensus that it should be removed. There was also a bug report filed around that time about this: 1274707. Having spent about a couple of days in the past week helping coworkers track down problems that ultimately proved to be obvious mistakes that were hidden by that catch clause, I've patched our local copy of boost 1.33.1 to remove that "feature". (I'd rather not think about how much time the affected coworkers lost to this.) I also checked boost CVS and found that this change hasn't been made there yet, so I'm attaching the patch below. ----------------------------------- --- thread.cpp.orig 2005-07-14 11:46:50.000000000 -0400 +++ thread.cpp 2006-06-16 18:17:33.000000000 -0400 @@ -105,22 +105,13 @@ static OSStatus thread_proxy(void* param) #endif { - try - { - thread_param* p = static_cast<thread_param*>(param); - boost::function0<void> threadfunc = p->m_threadfunc; - p->started(); - threadfunc(); + thread_param* p = static_cast<thread_param*>(param); + boost::function0<void> threadfunc = p->m_threadfunc; + p->started(); + threadfunc(); #if defined(BOOST_HAS_WINTHREADS) - on_thread_exit(); + on_thread_exit(); #endif - } - catch (...) - { -#if defined(BOOST_HAS_WINTHREADS) - on_thread_exit(); -#endif - } #if defined(BOOST_HAS_MPTASKS) ::boost::detail::thread_cleanup(); #endif