Unix Timestamp in Filepro

Fairlight fairlite at fairlite.com
Fri Oct 7 11:43:54 PDT 2005


On Fri, Oct 07, 2005 at 12:54:11PM -0400, Ryan Powers may or may not have
proven themselves an utter git by pronouncing:
> On Fri, Oct 07, 2005 at 12:16:42PM -0400, Fairlight said:
> 
> > *nix timestamps only have a one-second granularity.  If you're going
> > more granular than that, it's not the general *nix timestamp as measured
> > by seconds into the epoch.  There may be some extended format in some
> > software, but it's not the *nix timestamp standard as generated by time(2).
> 
> There is a difference between the "standard" *nix timestamp and what a
> system is capable of at the kernel level. To be sure I'd look at the
> timeval struct (or similar) in time.h. This struct contains the number
> of seconds into the epoch and also (typically) a 32-bit word for
> nanoseconds. You might not find it on a SCO machine but its been on BSD 
> for ages.

Oh, I know it can go down to nanoseconds at the kernel level.  Using ping
to something very fast and near will show you that much.  Ditto with
select().  The issue isn't whether the system can do it.

I'm just saying that asking for that granularity and specifying "unix
timestamp" are two entirely different issues, as one does not record to the
maximum granularity of the other.

Therefore, it's prudent to be precise when asking for something.

mark->
-- 
There is no "I" in TEAM.
This would be the primary reason I've chosen not to join one.


More information about the Filepro-list mailing list