3 ms·
There's a real chance the year 2038 problem will be worse than the year 2000 problem. Huge numbers of embedded things out there with no hope of software update/
by walrus01 4y ago
There's a real chance the year 2038 problem will be worse than the year 2000 problem. Huge numbers of embedded things out there with no hope of software update/new OS/patch that will do something weird in 2038.
- hn_throwaway_99 4y agoYou think 2038 will be bad, just wait for those poor saps that have to fix stuff 292 billion years from now when 64 bit integers overflow!
- Beltalowda 4y agoBy this time humans have evolved to be pure energy creates so we won't need computers any more.
- walrus01 4y agoDyson sphere problems
- codethief 4y agoExactly, consider how much legacy software will have been written by then and how they will need to patch all of it!
- Aperocky 4y agoSolved problem, most of those things will not last till then. The ones which does will be able to continue doing whatever it does in Dec 1901.
- phoehne 4y agoYou'd be surprised. PDP-11s and emulated VAXen are not for hobbyists. They're for companies that still have that PDP-11 or critical VAX they can't get rid of. It's usually in an embedded or context - like running a nuclear reactor. Sometimes things last a long-long-long time. (see https://logical-co.com/nupdpq/ https://logical-co.com/nupdpq/ or https://www.stromasys.com/solution/charon-vax https://www.stromasys.com/solution/charon-vax).
- phaedrus 4y agoI work on legacy software for an embedded system that supports transportation infrastructure. I'm pretty sure it has a year 2038 problem (based on first-principles; I haven't confirmed that with a test). The 32-bit time is baked into a serial data format and at the beginning of every internal device file. I don't think this is on anyone else's radar where I work. The only other person who spelunked in the software at this level retired at the end of 2022. (For better or worse I don't retire until after 2040.) I anticipate this is going to occupy part of my career just like Y2K did for a generation of programmers before me. That is, it will if and when I can convince management to prioritize it. Cynically I wonder if they may run out the clock, playing chicken with the last possible moment to begin to move on it. Our manager likes to say, "political, fiscal, technical - in that order" as an amoral description of how things operate. But this is different than refusing to implement, e.g. IPv6 by an artificial deadline.
- actionfromafar 4y ago32 bit time. It’s what plants crave!
- walrus01 4y agoWelcome to Amazon, I love you
- travisgriggs 4y ago> Our manager likes to say, "political, fiscal, technical - in that order" If I worked for your manager, my mantra would be “get out of her fast, network like mad, maximize my personal revenue —- in that order.”
- thunderbong 4y agoThat's a fact in practically every organisation. Just try to be part of organizations where the political element plays a lesser role. It will still be priority one, just not frequently enough to rankle.
- 4y ago
- bee_rider 4y agoIt’s kind of funny to think how few computing devices we had in 2000. I think we might have had… like 6? If you are really generous. Family computer, dad’s laptop (IT), probably a console, and a game boy, cable box, modem… I dunno, probably missed some. There are so many IOT devices in my place. Hopefully they’ll all just be replaced by 2038. Infrastructure will be a much bigger problem, but it belongs to somebody else!
- jillesvangurp 4y agoBut how critical is it that all these systems have the correct time? Mostly, this won't be that big of a deal. So what if some ancient device thinks it's 1970? Most of these devices don't even have UIs to set the time correctly: so they might have never had a correct notion of time to begin with. And how many of these devices have an uptime of 68 years? What happens when you turn these things off and on again? Do they start out at 1970? If not, where do they get the correct time? Embedded devices don't come with internet connections so there is no ntp running on them. Probably banks and insurers with ancient software might have to patch a few things up again. They would care about having a correct time representation. But they still have 15 years to figure that out. And at this point, you'd hope they'd be aware of this at least.
- Kamq 4y agoMy instinct says the IoT things are mostly fine. Might be some weirdness right around the overflow that makes things segfault (especially a time range crossing the event), but they'll probably just reboot. Banking and government/military do seem like the main pain point. Military specifically seems like it needs the most care because of the potential damage of something going wrong.
- duped 4y agoThere are two questions to ask. What happens if the date overflows? For example, if my toaster clock is wrong, do I care? How many of these devices are designed to last 16+ years? Because what I suspect will happen is that things that are pointless to patch won't be patched and the things that need to last 5-10 years will start shipping with fixed firmware in 2028-2033. Or more likely, late 2037.
- tiagod 4y agoFor anything that communicates over the Internet, the biggest problem I can think of is that wrong dates break TLS.