From: leon breedt Date: 2004-09-28T10:41:57+09:00 Subject: Re: safety of timeout() On Tue, 28 Sep 2004 08:52:33 +0900, Yukihiro Matsumoto wrote: > Define 'safety' first. Interpreter should not dump core by this > usage, but no exact 30 seconds guarantee. i'm not too worried about the precision, more just concerned about cases that can interrupt the timeout. like when a signal is received from OS while inside blocking code wrapped by timeout(). how does the scope of trap() work for these situations? does trap() work at the level of the real process? or current virtual thread?