Early 90s, NT 3.1 was released in 93. By release date, nearly 31 years old. First version that a significant number of people actually used was 4.0, 96, and the first consumers used was 2000. Development started in 89, as OS/2 3.0, then still a joint project with IBM.
For comparison, the first public Linux version dates to 91, 1.0 to 94, though Unix of course goes back to 69. Practically the only other thing still surviving from that era are IBM mainframes. In CS terms both are prehistorical, you can tell by how papers from back then aren’t typeset in TeX, worse, are usually scans. Yet they somehow had it all pretty much already figured out and we’re now often re-discovering insights that they simply didn’t have the hardware to implement back then. Not impossible to implement back then but here’s a fun one. And a modern walk-through through the thing.
And itself based on VMS, which was released in 1977.
Almost everything interesting about mass market operating systems was done in the 70s. Tons of academic work out there otherwise, but we’d have to rewrite everything to make good use of a lot of it.
L4 runs on billions of devices but it’s mobile modems, car infotainment, and stuff. There’s also been innovation generally in the mobile and restricted space, no desktop OS can manage application lifetimes quite as well as Android can: Because it needs to because you want both multitasking and fit everything into limited RAM.
The problem is really the “rewrite everything” part because to make current OSs much better you’d need to work on the interface between programs and OS, and not just piecemeal stuff like wayland. It was possible on mobile because everything was new, anyway, but on desktop? Truth be told our best bet at new standard APIs is wasm.
Um actuwally modern Windows is based off the Windows NT kernel that isn’t 30 years old 🤓Still from the early 2000s though so it’s no excuse.Edit: I am spreading misinformation online :3 I was thinking of its mainline Windows release.
Early 90s, NT 3.1 was released in 93. By release date, nearly 31 years old. First version that a significant number of people actually used was 4.0, 96, and the first consumers used was 2000. Development started in 89, as OS/2 3.0, then still a joint project with IBM.
For comparison, the first public Linux version dates to 91, 1.0 to 94, though Unix of course goes back to 69. Practically the only other thing still surviving from that era are IBM mainframes. In CS terms both are prehistorical, you can tell by how papers from back then aren’t typeset in TeX, worse, are usually scans. Yet they somehow had it all pretty much already figured out and we’re now often re-discovering insights that they simply didn’t have the hardware to implement back then. Not impossible to implement back then but here’s a fun one. And a modern walk-through through the thing.
I’ve always had a thing for older, more mature kernels.
KILF Kernels I would like to fuck
NT 3.1 was released in 1993. But sure, the kernel has been reworked a lot since then.
And itself based on VMS, which was released in 1977.
Almost everything interesting about mass market operating systems was done in the 70s. Tons of academic work out there otherwise, but we’d have to rewrite everything to make good use of a lot of it.
L4 runs on billions of devices but it’s mobile modems, car infotainment, and stuff. There’s also been innovation generally in the mobile and restricted space, no desktop OS can manage application lifetimes quite as well as Android can: Because it needs to because you want both multitasking and fit everything into limited RAM.
The problem is really the “rewrite everything” part because to make current OSs much better you’d need to work on the interface between programs and OS, and not just piecemeal stuff like wayland. It was possible on mobile because everything was new, anyway, but on desktop? Truth be told our best bet at new standard APIs is wasm.