On 03/16/2013 03:48 PM, g...@gabegold.com wrote:
My first year-end retirement account statement -- for 1971! -- listed my
projected retirement date as the incredibly distant, unimaginable,
science-fiction-like date of February 1, 2012. So back then at least TIAA-CREF
understood time windows spanning entire careers and beyond.
Quoting:
I was somewhat surprised that it took so long for Y2K to rise to the top, with 30-year
mortgages and such forward-looking things being far from uncommon. But I guess those
systems all got fixed one at a time as their windows passed 2000, without enough
awareness that anyone said "Hey, we need to fix this globally, now that storage
(both memory and DASD) is cheap!"
...
I think it more likely that those dealing with events of three-decade or
longer duration were already close enough to year 2000 when application
development took off after the 1960's to understand that applications
dealing with those events had to address century issues. Those dealing
with short-term events and with the experience of pre-S/360 early
computing, where hardware upgrades implied application re-writes every
several years, were understandably more likely to take a short-term
view. What many didn't foresee in the early days was that a large
volume of short-term data, application interdependences, and upward
compatible hardware would eventually preclude there being any
"convenient" time to address four-digit year issues before it became
mandatory to do so.
All the focus on early memory prices also glosses over the fact that
when the S/360 was introduced in 1964 even the largest models had a
physical memory limit of 8 MiB, and the S/360 architectural limit was
only 16 MiB. If you used a memory-extravagant design and created an
application requiring memory beyond available limits, the cost of
memory became effectively infinite - not available at any price. Larger
record sizes impacted requirements for not only main memory but external
storage capacity and channel bandwidth and processing time to manipulate
the data. There definitely was a mindset in place in earlier days that
placed great value on conserving all hardware resources as much as
possible, to the point of obsession in some cases. If you exceeded
available resources then, the hardware upgrade cost was more likely to
be cost-prohibitive, or the upgrade might not even be possible.
With 20/20 hindsight we would have liked some past design choices to
have been different. Hardware got relatively cheap over the years and
changed the design equation, but the cost of people-hours to re-work old
code designs replaced hardware as the limiting factor.
--
Joel C. Ewing, Bentonville, arjcew...@acm.org
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN