This has come up a lot from some of my clients as possible problem dates as
well.  I agree with your thinking Paul, but I was able to come up with a
scenario.  If they were using a 16 bit integer to store the data, and it
was some bit representation like CALENDAR or something, and they were using
COBOL with a PIC S9(4) COMP, then the largest value you could get in would
be 9999.

I don't know anyone that has actually done this, but I do know that a
number of ManMan shops just went through what they were calling "The ides
of march".  I guess the original ASK Date format was similar to Calendar in
that it was a julian date that counted from the formation of the company.
The way they implemented it the field maxed out on March 15th of this year.
CA fixed this a while back, but no everyone got update.

Shawn





Paul Christensen <[log in to unmask]> on 03/31/99 06:05:08 AM

Please respond to [log in to unmask]

To:   [log in to unmask]
cc:    (bcc: Shawn Gordon/IS/FHM/FHS)
Subject:  Y2K date problems




I have read a similar story about Y2K date problems from several sources
(Computer World,
CNN.COM and a Twin Cities newspaper), and I fail to see why their story is
not just some
misinformation.

The story is that April 9th 1999 and September 9th 1999 are two possible
dates for problems.
The reasoning is that both dates (one Julian and one Gregorian format)
cause a date of 9999,
which many programmers may have used to signify end-of-file.

Now I understand the concept of using all 9's to signal end of data in a
file.  In the past, I too
have used that method.   And the only reason I haven't lately is not that I
got any smarter about
using dates, just that most programs I write today are on-line database
updates, and this
technique I used was in batch programs where I had to merge two or more
data files.

But to me, all 9's would mean six 9's (999999), or 5 9's in the case of
Julian dates; but never four nines.   And besides, who stores dates without
the leading zeroes?
   9/9/99 would be stored as either zero nine, zero nine, ninety nine or in
YMD format of 990909.      September 9th does not get stored as 9999 in any
computer system that I know of.
 Nor would a Julian date of 99099 get stored as 9999 either.
You have to store the zero in order to know the significance of the
numbers.

Now maybe somebody here can point out  the error in my thinking, but my
feeling is these
two dates are completely off the mark as far as being a problem date.  And
there are enough
real problems with Y2K, we don't need some imaginary ones!

Paul D. Christensen
PC Enterprises Inc.
Osakis MN