alarm (was Re: Linux-1.2.13 and Linux-1.3.15 )
Marty Leisner (leisner@sdsp.mc.xerox.com)
Mon, 7 Aug 1995 07:40:42 PDT
> "Marty Leisner": "Re: Linux-1.2.13 and Linux-1.3.15" (Aug 4, 13:19):
> > > - some timer fixes ("alarm(-1)" works reasonably, itimers check for
> > > overflow)
> >
> > alarm is unsigned int..(accordling to the gnu libc
> > manual...
> >
> > What does reasonable mean?
>
> Reasonable means that "alarm(-1)" will result in a _long_ timeout. It
> won't actually be equal to (~0) seconds, as the internal timer
> resolution on the i386 is a 100Hz 32-bit value, but it will be around 20
> million seconds in any case, which is "good enough" in my opinion.
>
hmmm...why isn't alarm(-1) == alarm(~0)
Mentally, 32 bit ~0 seems to be around 20 million seconds....well, whatever,
its enough...
Actually, any alarm(<0) should delay for a long time...
marty
leisner@sdsp.mc.xerox.com
Member of the League for Programming Freedom