Hacker News Reader

Comment by reaperducer | Hacker News Reader

Nobody ever gets credit for fixing problems that never happened (2001) [pdf]

I'm reminded of this every time I see some YouTuber or other social media click bait claiming that the Y2K bug was no big deal.

The reason it was no big deal is because thousands of graybeards, like myself, stayed up many long nights for months ahead of time making sure things would work.

I still remember the tension during the countdown to midnight UTC. Then the tension rising again during the countdown to Eastern time. Then once more at local time.

Only when it was 2000 in Pacific did we unclench our cheeks.

View on HN

Depending on which Y2K bug you're referring to, this very much happened and is an easily believable scenario. I don't think this is a comparable situation.

I still don't quite buy that. Surely it's because computers mostly use epoch time for dates, not dd/mm/yy?

Guess we'll find out in 2038.

Now they do at least

They did then as well. I remember before 2000 people saying it wasn't going to be a big issue due to this.

"Most" programs used epoch time, like video games. The minority that didn't were the big, stable, old programs that ran on big computers. Things like payroll and flight schedules.

Many old programs written last century did not use epoch time.

https://en.wikipedia.org/wiki/Year_2000_problem#On_1_January...

It's worth noting that most of the programs affected would not be the ones people used on their private pc.

That's not what I remember programs looking like. You would have counters that went to 99 and no more because the UI wouldn't accommodate another digit, and programmers didn't know what they were doing, and figured it would be clearer if the software reflected its UI and couldn't handle over 99 at all, invalid UI means unrepresentable state.

Lots of stuff didn't care about the day, too, like you'd get 0997 and that meant September 1st 1997 because the first (or last) of the month was inferred, and you'd get goofy logic around year++ where you add 100 years whenever you want to increment the month, and the whole thing is written as a modulo 1200 but whenever the date is about to be 0000 you look at what's stored in year instead and add one to it instead of 100, because that way you have less variables to allocate and every "little bit" counts.

Everyone knows now, but lots of people writing programs didn't have any prior art, they were just good at messing with computers and sort of fell into programming by accident.

Interesting that everyone is responding to the first half of your comment. I think 2038 is going to be more interesting than Y2K...

https://en.wikipedia.org/wiki/Year_2038_problem

It was also to do with how dates were stored in very old databases that were conceived in the 70s when storage was at a premium - a lot of them just used two digits for year.

Reply on news.ycombinator.com