Very Computer

by **Mark A Harris** » Fri, 27 Sep 1996 04:00:00

Is there any guarantee that gettimeofday will set the tv_sec
field to be 0 <= tv_sec < 1,000,000 ?

In other words, for struct timeval's returned by gettimeofday, is
it safe to assume that if a.tv_sec > b.tv_sec then a > b ?

Thanks
Mark.

--
Mark Harrison http://jasper.ora.com/mh/

by **Jeff Bonwi** » Fri, 27 Sep 1996 04:00:00

Quote:>Is there any guarantee that gettimeofday will set the tv_sec
>field to be 0 <= tv_sec < 1,000,000 ?

I assume you meant tv_usec, in which case, yes -- you can assume that
the number of microseconds is in the range 0 <= usec < 1,000,000.
In general you can assume that any system service that returns a timeval
(second and microseconds) or a timestruc (seconds and nanoseconds)
will put the fractional part in the range 0 <= fraction < 1 second.

Jeff Bonwick
Solaris Performance

by **Casper H.S. Dik - Network Security Engine** » Wed, 02 Oct 1996 04:00:00

Quote:>Is there any guarantee that gettimeofday will set the tv_sec
>field to be 0 <= tv_sec < 1,000,000 ?
>In other words, for struct timeval's returned by gettimeofday, is
>it safe to assume that if a.tv_sec > b.tv_sec then a > b ?

Yes, the tv_usec value should be < 1000000; if not, it's a bug.

(Though such bugs did exist in the past)

Casper
--
Expressed in this posting are my opinions. They are in no way related
to opinions held by my employer, Sun Microsystems.
Statements on Sun products included here are not gospel and may
be fiction rather than truth.

Very Computer

range guarantee on gettimeofday?

range guarantee on gettimeofday?

range guarantee on gettimeofday?

range guarantee on gettimeofday?