Le 29/07/13 14:44, Klaim - Joël Lamotte a écrit :
As someone suggested to me in the comments of the page, I re-implemented the loop using a condition_variable so that pushing a task in the queue notify it.
I have a friend who can reproduce the bug systematically (windows Vista, core 2 duo) so I sent him versions:
- removing the sleep call works (but is a bit less performant apparently) - changing the loop to use condition_variable works
So it looks like the sleep call does have a strange behaviour but I don't know the source of the problem. Using the condition_variable seems more appropriate so I'll keep this fix for now.
Hi, there were an issue fixed on 1.54 https://svn.boost.org/trac/boost/ticket/8136. Please, could you tell me if this fix is related? Best, Vicente