Re: 1/1/1970

New Message Reply About this list Date view Thread view Subject view Author view Attachment view

From: Mike Eisler (mre@jurassic.eng.sun.com)
Date: 03/18/97-04:05:22 PM Z


Date: Tue, 18 Mar 1997 14:05:22 -0800 (PST)
From: Mike Eisler <mre@jurassic.eng.sun.com>
Subject: Re: 1/1/1970 
Message-ID: <Roam 0.9.4.858722722.13317.mre@jurassic.eng.sun.com>

> 
>    >> If 2106 is too soon, then why not change the format from two 32 bit
> value
>   s
>    >> to a single 64 unsigned value.  With a 1970 base, this then ranges out
> to
>    >> around 2554.
>    >
>    >Interesting idea, though doesn't this burden OS's or filesystems
>    >that do not support time resolutions below 1 second ?  Currently
>    >they can just plug in the seconds value and set the nanoseconds
>    >to zero.  If the entire value is represented as 64 bit nanoseconds
>    >then they'll required to do some non-trivial 64 bit arithmetic.
> 
> Another possibility (that doesn't even need to be part of NFS V4) is to
> store the high two bits of a 34 bit seconds value in the high two bits of
> the nanoseconds.  Reassembling the data would be faster than the divide
> Lance is suggesting.  Of course, no one would bother to implement it until
> the year 2100.  (Which, of course, is *not* a leap year, but that's a topic
> for a different forum.)

I like this. We could use one of the spare bits as a sign bit giving us an
approximate range of 1697 - 2242 A.D. Not good enough for physicists that want
to use O/S file times, but certainly adequate for commodity futures and banks
and some historical applications.

As Rick says, no one would bother to implement it for some# of years (I would
hope someone would try it some time before 2038), but it gives our successors
plenty of time to figure out what V5 should be.

	-mre


New Message Reply About this list Date view Thread view Subject view Author view Attachment view

This archive was generated by hypermail 2.1.2 : 03/04/05-01:45:33 AM Z CST