Subject: Re: On-disk time_t
To: Bill Sommerfeld <>
From: Bill Studenmund <>
List: tech-kern
Date: 01/03/2000 11:12:55
On Fri, 31 Dec 1999, Bill Sommerfeld wrote:

> It occurred to me at recently that if you don't have a real need for
> nanosecond-granularity file timestamps (and can make due with, say,
> microsecond granularity), you can steal the 12 high-order bits out of
> the 32-bit word containing the "microseconds" value, which would get
> you to 2**44 seconds after 1/1/1970, which, assuming i did the math
> correctly, gets us to well past the Y100K point.

Using unsigned values, we get to past 559K with a 44/20 split such as the
one you describe above. :-) With a signed value, we only get to around

Take care,