I've read a lot online about how "Y2K was overblown". As an engineer who worked like crazy in 1999 to correct our systems this drives me crazy.
Today I read about games going offline, watches getting frozen and airline tickets showing the wrong date.
Because of a leap year. Which happens every FOUR years (give or take).
Yeah Y2K was a one time thing. (Well at least until year 10000)
Isn't 2038 going to be fun.
@jonathankoren Or below one. Or sufficiently close to any plane's probable trajectory. Especially Boeing planes.
You have my sympathy. I get incredibly annoyed with fuckwits who claim Y2K was a big nothingburger, after many many people worked their asses off to make sure it didn't crash civilization as we know it. I'm like "those bridges you drove over today, did they fall out from underneath you? No? Oh, does that mean the engineers overspecced them?"
@rhempel I'm more annoyed we use data types that are for the computers convenience rather than the humans and systems themselves.
COBOL got this right FWIW.
@mirabilos @rhempel But it was plainly obvious that's what it was doing. Dates were a bit weak but numbers and strings were great.
PIC S999,999,999V99 clearly indicates what the maximum and minimum numbers are :) Additionally you don't get unexpected IEEE floating point math rounding issues.
@damieng @JorgeStolfi @dschaub @rhempel Any breakage from the size of `int` changing is a sign of code written with assumptions that were poor in 1989 when the language was first standardised.
Granted this whole thread is about the fallout from poor assumptions, but I'd prefer to assume "surely the code will have been recompiled for a wider target/updated for greater storage by then" than "surely the Standard's explicit allowance for wider targets will have been deliberately ignored".
@JorgeStolfi @damieng @rhempel @dschaub @mlp that would have broken anything using u_int32_t as that type would suddenly be promoted to signed int in arithmetics, plus we need a type in between short and 64-bit.
LP64 was the most sensible choice. LLP64 (as in NT) would not have worked with the vast majority of Unix code that puts those types as long, which was found in both earlier and later code than the code putting them into int.
The downside of that change is that the "%d" and "%ld" formats still require "int" and "long int" values. So in order to print an int64_t value I may need one of the other, depending on the platform. I must use some ugly hacks to make my code portable...
The "%f" format can handle single and double floats. Why can't "%d" handle ints of ANY size?
@damieng It really is pathetic to see this happen today.
Nothing was seemingly learned from Y2K.
@marcmagnin Top of that page though "Turn your computer off.." completely misses the point.
You can't patch anything with it turned off and both Y2K and 2038 are still an active issue when you turn it on days later (unlike leap seconds/leap years)
@hengymrohebwlad @7faces @damieng It was using interval. A cautionary tale of code that can run for years without issue.
This may exist and I just don't know about it with Oracle, errors like this seem like a good fit for something static analysis could weed out.
@7faces @damieng I worked on Oracle for over 20 years and never used "interval". I also don't recall ever hitting a leap year problem.
Now I work on Teradata, I see lots of interval arithmetic in other people's code. And suddenly I hit a leap year problem.
Coincidence? 🤔
I'll change these intervals to ADD_MONTHS if possible when I stumble across them and see what happens.
What could possibly go wrong? 😱
@damieng there's a specific variant of problem (or maybe it's all problems?) that is hamstrung in the public eye because to a layperson, a situation wherein one is correctly prepared for it makes it look like it wasn't a serious problem in the first place.
covid, y2k, climate change come to mind as examples of how the problem manifests with different levels of preparedness. and i genuinely have no idea how to get some people to care about these kinds of issues.
you misunderstand me, i'm saying that if covid were correctly prepared for, you'd have vast swathes of people saying "Why did we go to such extreme measures to deal with something that wasn't even that bad in the first place?"
they are the same category of problem but y2k was actually prepared for, whereas covid and climate change were and are not.
@damieng @negativeprimes "Y2K was overblown" <= tell me you know almost nothing about computing without telling me you know almost nothing about computing
The fact that people think it was overblown at all is a testament to the outrageous amount of work people did to prepare for what could have been a very unstable time
Chris E
Damien Guard
Frank Gilsdorf
hairylarry
it's B! Cavello 🐝
Kevin Dente
Alexander Janßen
Nefarious Celt
jonathankoren™
Wilfried Klaebe
·J Mopp
Kevin Lyda
Chris Gerhard 
Artemesia
Justin M
rhempel
DavidSchaub
Jorge Stolfi
mlp
mirabilos
Dave Irving
Magical Cat
Ambient Sponge
jimbodon
Erin🏳️🌈🏳️⚧️🍻🖖☘️🦖💥🍪
Marc Magnin
Robert Lou Dobbs III
Hen Gymro Heb Wlad
Bubba Harrison
Simonoid
ivory
Mark Fischer
PhilSalkie
trenchworms
iwein
refraction 
neatchee