really needs a fractional second?) I'll also clone the bug over to libc for a better fix, as I think it really needs to be handled there because trying to work around the problem means making assumptions about the implementation that I'm not sure are/will be valid.
I already thought about possible fix in libc side. The problem is that the precision of absolute time and relative time is the same. What should we do with one-shot timer, when now+delay > max(time_t) ? And what should we do with repeating timer, when now+10*delay > max(time_t)I.e. man setitimer says "On Linux kernels before 2.6.16 timer is silently truncated to this ceiling value".
Should we do the same with timer_settime() ? Petr