Email or username:

Password:

Forgot your password?
Csepp 🌢

Trying to figure out the architecture for a somewhat self-hosting OS for the #PocketViewer. Instead of sleeping.
#CollapseOS would only give me access to about 28K of near addressable RAM, but maybe that would be enough?
I don't see much point in scrapping the existing OS though and its standard library would be nice to have access to. Well... I guess I could start by shipping cOS as a C application and providing a C call primitive...?

4 comments
Csepp 🌢

If I could take over the OS then I'd get 128K of RAM, that's two full 16 bit segments! Wow.

Csepp 🌢

Also still thinking about a type system for Forth... probably wouldn't make sense for cOS, but for Dusk OS and my DS OS dream project... it might just work.
One route would be to take existing cOS/dOS code and verify its correctness with external annotations, the other would be something more like Factor, where the type information is also used for interface resolution and optimization.
I guesss the second could be built on the first....?

Devine Lu Linvega

@csepp I've recently implemented factor's stack effect inference and it's a ton of fun to watch it find stack imbalances, extending forth to do this sort of preprocessing gymnastics is a blast :)

I'm looking forward to see what you cook up!

docs.factorcode.org/content/ar

Csepp 🌢

@neauoire I'm kind of mentally stuck on typing READ, because then TOS determines stack depth. Could certainly do it with a complex dependent type system, but then I probably won't have much code space for anything else.

Go Up