Data Archival (OT Long)

From: Jerome Fine <jhfine_at_idirect.com>
Date: Tue Dec 12 08:27:30 2000

>Sellam Ismail wrote:

> On Sun, 10 Dec 2000, Jerome Fine wrote:
> > The only question that I still have is the format used to display the date
> > when the year is greater than 9999. Any suggestions?
> DD-MMM-YYYY(Y...)

Jerome Fine replies:

I find that reading 31-Dec-99,999 is preferred to 31-Dec-99999, but
I don't know what the standard will be. Or perhaps 99,999-Dec-31
since with other dates in a column for a directory listing, I suspect
that 31-Dec-08,765 might seem a bit strange rather than
FOO.BAA 8,765-Dec-31
FOO.BAB 28,765-Dec-31
FOO.BAC -2,345-Dec-31
But, if the user is allowed to choose the date format, that would be
even better. Then for dates between 1000 and 9999, the user can
still have his cake and eat it at the same time.

> But you do realize it will be 8,000 years before anyone cares, right?
> This period of time encompasses most (if not all) of our current recorded
> history.

Then why do we look at Egyptian literature, some of which is now 5000
years old. I realize that not everyone is interested, but if the work is being
done to make the programs Y2K compliant and it takes only double the
effort to make them Y10K compliant, it was that same attitude that led
to the Y2K problems in the first place. What I am saying is that getting
the software to work until 2099 is very easy since there are no technical
or format problems involved. But, extending the date to even 4000 means
solving a few questions like where to put the extra year bits. However,
once those questions are solved along with allowing for minor hiccups
like which years in the Common Era (Gregorian) calendar are no longer
leap years (or in some cases there might need to be extra leap years),
setting up a reasonable display format for the date with more than 4 year
digits is actually the most difficult part of the coding. And if someone
does decide to actually run the software under multiple emulators, then
imagine the surprise when the current date is actually working - assuming
that the current Common Era calendar is even understood.

> By that time everyone will have left the planet. Our computers will be
> all alone and lonely.

I doubt that we will all have left the planet. The alternative is that we will
have bombed or otherwise gutted ourselves back to the stone age. Will
that then qualify as a topic that is over ten years old? Like going back to
the future?

Seriously, however, the point I am trying to make is that the technical
problems that need to be solved are the same whether the date is extended
to 31-Dec-2999 or 999,999-Dec-31. Once there are a few extra bits
around to handle another 900 years, another million years is trivial except
for the display aspects. The internal bits used to record the date need
only 4 more bits for the year 2999, but it is just as easy the use 2 bytes
which then allows years up to at least 4,000,000 in the future. As I
understand the VMS time value, there is room to use dates up to about
the year 25,000 - so the only problem will be the display format for
a very long time - unless the code does not allow for leap years. Since
old versions of VMS are over 10 years old, I think this discussion is
still on topic - how long old code will still function - the same topic as
how long old hardware will still function.

Sincerely yours,

Jerome Fine
Received on Tue Dec 12 2000 - 08:27:30 GMT