On Tue, May 08, 2001, cvs at videolan.org wrote: > Henri: a 0.5s timeout is not (time_t)0.5 sec and (time_t)0 usec > but (time_t)0 sec and (time_t)500000 usec > Now the input doesn't eat all the cpu any more :p I owe you a beer :-) -- henri