Email or username:

Password:

Forgot your password?
Top-level
Royce Williams

@plaes

I see your point, though I must object in part - most of the IT people and programmers fixing the problem weren't the people who created it. I'd say that most of us "inherited" it. :D

Not that I strongly blame the original programmers, either - it emerged organically for understandable reasons.

@sinbad @robertatcara

10 comments
Jeff C. 🇺🇦

@tychotithonus @plaes @sinbad @robertatcara I wouldn’t blame those early programmers at all. They were largely working within the constraints of very limited systems.

I’d blame those in charge of budgets — bean-counters — who didn’t want to pay to refactor old systems, even decades after it was more feasible, to handle dates properly.

Y2K is past us, but the fight to refactor and maintain “unprofitable” code remains.

Billy Smith

@jeff @tychotithonus @plaes @sinbad @robertatcara

One engineering professor that taught us in the late 80's used to recommend occasionally doing something like the above cartoon, in business-critical, but non-safety-critical systems, as a way of reminding the people on the business-side of enterprises, why they should spend the money maintaining the infrastructure. :D

To quote him, "I don't like this approach, but it works." :D

IceNine

@catch56 @jeff @tychotithonus @plaes @sinbad @robertatcara

Good days to take off work:

* Feb 7, 2036: NTP 32-bit unsigned time stamp values will overflow

* Jan 19, 2038: Unix Time 32-bit signed value will overflow

* Nov 20, 2038: GPS week number rollover

* Dec 31, 2078: Excel Date 16-bit value will overflow

* Feb 7, 2106: Unix Time 32-bit unsigned value will overflow

Kjetil Kilhavn

@jeff @tychotithonus @plaes @sinbad @robertatcara One of my first colleagues said they were through the same thing when 1969 was about to turn into year '0'. One would initially think they'd have fixed it properly then, but satellite communication was expensive, so the fix was to add one digit to the year and fix it for the next 30 years. #y2k

ChrisR

@jeff As a very early coder, you are so right about those constraints. In 1968 or so I was supporting suites of structural analysis program that had to operate in a max memory of 12,000 24-bit words (48 Kbytes) with no disk. We did 250-node grid and planeframes, or 100-node 3D structures. Matrix inversion (by Newton's decomposition) between 2 tape drives was a sight to see! I had to include an operator-console countdown to convince the Operators (!) not to cancel the jobs for apparent looping.

Go Up