Picture this: it’s January 19th, 2038, at exactly 03:14:07 UTC. Somewhere in a data center, a Unix system quietly ticks over its internal clock counter one more time. But instead of moving fo…
I don’t understand why people always say that. Pentium Pro could handle 64 GB even though it was a 32 bit CPU. It had a 36 bit address bus. Later models are the same.
People say it because it was a Windows limitation, not a computing limitation. Windows Server had support for more, but for consumers, it wasn’t easily doable. I believe there’s modern workarounds though. The real limit is how much memory a single application can address at any given time.
What does a 64-bit system and 4GB RAM have to do with using 64bit timestamps?
32bit systems can use 64bit values without issue. In fact, even an 8bit system can handle 256bit values or even longer ones without issue.
The bittiness of a CPU and its address space have nothing to do with the length of usable data unless you end up with data longer than the RAM volume (and even then there’s swap).
There is also the fact that during Y2K, we didn’t have as much reliance on computers.
And we still shouldn’t.
Uniting the reliance upon long-range electric connectivity (radio, PSTN - but that now depends on computers too), the reliance upon computers (like mainframes), the reliance upon microcontrollers, the reliance upon personal computers (like Amiga 500), the reliance upon fast encryption helped by computers, the reliance upon computers used for mining cryptocoins or some beefy LLMs, the reliance upon computers capable of running Elite Dangerous, and the reliance upon computers capable of running devops clusters with hundreds of containers, - it’s wrong, these are all different.
An analog PSTN switching station shouldn’t care about dates. A transceiver generally shouldn’t too. A microcontroller doesn’t care which year it is, generally.
With an Amiga 500 one can find solutions, and it’s not too bad if you don’t.
The rest is honestly too architecturally fragile anyway and shouldn’t be relied upon as some perpetual service.
The problem doesn’t concern me as much at how bad we’ve become at maintaining shit that already works.
There is also the fact that during Y2K, we didn’t have as much reliance on computers.
There was also a worldwide effort to fix any potential problems before they happened.
Cobol mavens burned both ends of the candle and made bank, while making banks work.
Many were old enough to retire after that.
Issue 2038 will be easier to fix because many systems are already 64-bit, as 32-bit systems could only handle 4 GB of RAM, and programs need more RAM.
The only issue would be critical issues that run on 32-bit systems and must be fixed before that date.
So, many banks and government agencies which still run on mainframes…
I don’t understand why people always say that. Pentium Pro could handle 64 GB even though it was a 32 bit CPU. It had a 36 bit address bus. Later models are the same.
People say it because it was a Windows limitation, not a computing limitation. Windows Server had support for more, but for consumers, it wasn’t easily doable. I believe there’s modern workarounds though. The real limit is how much memory a single application can address at any given time.
What does a 64-bit system and 4GB RAM have to do with using 64bit timestamps?
32bit systems can use 64bit values without issue. In fact, even an 8bit system can handle 256bit values or even longer ones without issue.
The bittiness of a CPU and its address space have nothing to do with the length of usable data unless you end up with data longer than the RAM volume (and even then there’s swap).
And we still shouldn’t.
Uniting the reliance upon long-range electric connectivity (radio, PSTN - but that now depends on computers too), the reliance upon computers (like mainframes), the reliance upon microcontrollers, the reliance upon personal computers (like Amiga 500), the reliance upon fast encryption helped by computers, the reliance upon computers used for mining cryptocoins or some beefy LLMs, the reliance upon computers capable of running Elite Dangerous, and the reliance upon computers capable of running devops clusters with hundreds of containers, - it’s wrong, these are all different.
An analog PSTN switching station shouldn’t care about dates. A transceiver generally shouldn’t too. A microcontroller doesn’t care which year it is, generally.
With an Amiga 500 one can find solutions, and it’s not too bad if you don’t.
The rest is honestly too architecturally fragile anyway and shouldn’t be relied upon as some perpetual service.