Re: Y2K Fears to Stop Cityrail Trains

["Bill McNiven" <wmcniven@gunzel.ozemail.com.au>]

Michael McDonald wrote in message <7o8us3$5sc04$2@titan.xtra.co.nz>...
<much snipped>
>Digital computer systems had a potential problem of similar nature a couple
>of years ago - May 18, if I remember correctly,  but can't remember which
>year it was.  Date was significant because it was 10000 days since the base
>time of 1-Jan-1960 for Unix systems.


This is possibly perpetuating a subject that is not aus.rail, but I hope
that non-IT readers will excuse me as I atttempt to dispel some of the gloom
and doom.

1    A long time ago (ca. 1978) I tried to set standards in an electric
power utility for this.  Everything would be 32-bit days-since-31-Dec-1899.
This worked for a while, but new graduates who had learnt their own versions
of the 'unique truth' usually created their own standards, then ignored
them.
2.    My first experience of a HARD failure was in a system in which
maintenance cost transactions were written to a history file with a purge
date.  The customer had elected to keep details for 8 years.  All the
transactions for 01/01/1992 were kept until a date shortly after 01/01/1900
(goodbye, BHA MMS!).
3.    The significance of 18 May 1995 was that it was day 10,000 in PICK
(not necessarily DEC) systems, where the date is integer-days-since-31/12/67
(goodbye, TSW MPAC-UX! (Eagle Technology in NZ)).
4.    Besides the potential year 00 problem, there is also the 29th Feb
problem.  According to Tasmanian newspapers at the time, the Bell Bay
Aluminuim smelter was shut down on 29/02/1996 because their process control
systems (written in the 1960's) had a hard-coded list of leap years that
ended in 1992.  Certainly there was a shutdown.  The explanation is only
journalese from the Launceston Examiner and the Burnie Advocate at the time.
Apologies to those gunzels who see this as off-topic,
Regards,
Bill