On Dec 3, 2009, at 9:38 AM, Peter Dimov wrote:
Stonewall Ballard wrote:
I see that now. It's necessary for the interrupter to lock the condition variable mutex before interrupting the thread. The likelihood of failure would depend on the number of threads waiting on the condition variable, and the use of an interrupt_all() to interrupt them repeatedly.
So this really is a bug in boost::thread, at least with the pthreads implementation. I suppose that makes it easier to fix than a bug in pthreads.
...
I'm surprised that I can't find any other reports of this problem.
You might want to file a ticket for this issue on svn.boost.org so that it doesn't get lost.
Done. Ticket #3735. - Stoney -- Stonewall Ballard stoney@sb.org http://stoney.sb.org/