It’ll only affect 32bit systems with ancient operating systems storing dates in epoch time.
Not a small number. But nowhere remotely near what Y2K could have been.
Hopefully by the time we need to account for a 64bit rollover, I’ll be comfortably retired. But by that time, proton decay may be a more worrisome problem.
I know. I’m old enough that I worked through the Y2K problem. Not me literally - I was working on a different class of systems - but I literally sat next to COBOL devs who were paid to work on green screens on an IBM midframe for more than half their time to get rid of the two digit date representations on systems operating cellular communications as well as the ones that ran sales and services for a large telecom company. It was my first real job in the industry, and I remember the Gateway type computers sold at Sears with the “Y2K Compatible!” stickers on the front.
My phrasing was both tongue in cheek and a callback to another problem that similarly had some people dreading the end of the world with nuclear reactors running amok and planes crashing from the sky.
In any case, he had a bigger impact on the world than most humans ever will, and going out peacefully at 85 really doesn’t sound all that bad.
It would have just been really funny if his gravestone could have listed his dates as Born June 6 1936 - Died December 13 1901.
Most microcontrollers have only recently made the switch to 32bit, some are still 16bit. Particularly those used for IoT, where they might still need to check dates for TLS and other certs, could go belly up on 2038.
With 14 more years of IoT creep, it might as well have more impact than Y2K, like “sorry Dave, I can’t open the front door, your key won’t be valid for another 136 years”.
Most microcontrollers also don’t need wall clock, monotonous time is sufficient. uint32_max counting in seconds gets you 136 years of uptime before the thing wraps which exceeds lifetime, in milliseconds it’s 50 days which is well within the warranty window.
And, no, 64bit time_t is generally not an issue for 32-bit platforms: Those upper bits only get touched once every 136 years. OS vendors did step up and change stuff so everything that gets produced now should be fine. And don’t even start complaining about memory budget while those things are talking json.
in milliseconds it’s 50 days which is well within the warranty window.
LOL. Stuff like that reminds me how great we have it in the EU, where warranty is 2 years minimum. 🇪🇺😁
Anyway, I’d mostly worry about “really smart” electric engineers rolling out their own time and crypto functions from scratch, because “trust me, I’m an engineer”.
This is an actual OMG moment.
The next Y2K style problem will happen on this date, January 19, at pi o’clock in 2038. I was really hoping he’d get to see that.
He was ironically taken too soon.
It’ll only affect 32bit systems with ancient operating systems storing dates in epoch time.
Not a small number. But nowhere remotely near what Y2K could have been.
Hopefully by the time we need to account for a 64bit rollover, I’ll be comfortably retired. But by that time, proton decay may be a more worrisome problem.
I know. I’m old enough that I worked through the Y2K problem. Not me literally - I was working on a different class of systems - but I literally sat next to COBOL devs who were paid to work on green screens on an IBM midframe for more than half their time to get rid of the two digit date representations on systems operating cellular communications as well as the ones that ran sales and services for a large telecom company. It was my first real job in the industry, and I remember the Gateway type computers sold at Sears with the “Y2K Compatible!” stickers on the front.
My phrasing was both tongue in cheek and a callback to another problem that similarly had some people dreading the end of the world with nuclear reactors running amok and planes crashing from the sky.
In any case, he had a bigger impact on the world than most humans ever will, and going out peacefully at 85 really doesn’t sound all that bad.
It would have just been really funny if his gravestone could have listed his dates as Born June 6 1936 - Died December 13 1901.
Most microcontrollers have only recently made the switch to 32bit, some are still 16bit. Particularly those used for IoT, where they might still need to check dates for TLS and other certs, could go belly up on 2038.
With 14 more years of IoT creep, it might as well have more impact than Y2K, like “sorry Dave, I can’t open the front door, your key won’t be valid for another 136 years”.
Most microcontrollers also don’t need wall clock, monotonous time is sufficient. uint32_max counting in seconds gets you 136 years of uptime before the thing wraps which exceeds lifetime, in milliseconds it’s 50 days which is well within the warranty window.
And, no, 64bit time_t is generally not an issue for 32-bit platforms: Those upper bits only get touched once every 136 years. OS vendors did step up and change stuff so everything that gets produced now should be fine. And don’t even start complaining about memory budget while those things are talking json.
LOL. Stuff like that reminds me how great we have it in the EU, where warranty is 2 years minimum. 🇪🇺😁
Anyway, I’d mostly worry about “really smart” electric engineers rolling out their own time and crypto functions from scratch, because “trust me, I’m an engineer”.