Agreed! It's on the roadmap. The tricky bit is that the current version of WASI is synchronous, so the only way to integrate it with WANIX or anything that isn't running in the same worker is to use SharedArrayBuffer, which adds complexity and extra security constraints. That said I have a prototype of the mechanisms needed, so it's possible even before WASI 0.3 (I think) introduces async. Would love people with motivated use cases to help us prioritize it though.
This is really cool stuff. I am having a tiny bit of trouble seeing the use case but I think thats just because its early. The tech here though is epic.
Really cool BUT if you are starting from scratch and want to take inspiration from stuff like smalltalk, why not listen to Alen Kay and not try to repeat the same 50 years old Unix design?
Thanks! Yes, there is plenty we could leave behind. However, I've noticed that without grounding in something familiar, you can make something too "out there" to be practical/compatible/understandable. Also, instead of Unix/Linux, I'm pulling more from Plan 9, which had great refinements of Unix ideas that have not totally made their way outside of that research context. I think the real trick, though, is taking this familiar paradigm and pushing into being something that is actually totally new without you realizing. We'll see how it plays out. Hope you'll be following along!
That's....really boring! It's just another iteration of "the thing we could already do, but in the browser." Jeff even says, "there are no VMs", and then has to clarify, "just WASM"(a VM). Then they share the screens and do the CRDT thing and he goes, "it's like we both SSH'd into a shared server, except there's no server!" But there IS a server - it's the one that's exposed by the service worker! Huh, yeah, it's almost as if we should go back and fix the UX around the actual OS instead of doing whatever this is.
Supporting WASI would be rad
Agreed! It's on the roadmap. The tricky bit is that the current version of WASI is synchronous, so the only way to integrate it with WANIX or anything that isn't running in the same worker is to use SharedArrayBuffer, which adds complexity and extra security constraints. That said I have a prototype of the mechanisms needed, so it's possible even before WASI 0.3 (I think) introduces async. Would love people with motivated use cases to help us prioritize it though.
WASM ❤
This is next level
This is really cool stuff. I am having a tiny bit of trouble seeing the use case but I think thats just because its early. The tech here though is epic.
Cool presentation!
I love wasm
awesome. Mozart's Ghost! The hottest band on the internet!
Pretty rad!
Can we get the slides?
Guys, please reduce the volume of the intro! Otherwise cool presentation, thanks!
Fuckin hell.
Not WAOS :( xP
Really cool BUT if you are starting from scratch and want to take inspiration from stuff like smalltalk, why not listen to Alen Kay and not try to repeat the same 50 years old Unix design?
Thanks! Yes, there is plenty we could leave behind. However, I've noticed that without grounding in something familiar, you can make something too "out there" to be practical/compatible/understandable. Also, instead of Unix/Linux, I'm pulling more from Plan 9, which had great refinements of Unix ideas that have not totally made their way outside of that research context. I think the real trick, though, is taking this familiar paradigm and pushing into being something that is actually totally new without you realizing. We'll see how it plays out. Hope you'll be following along!
That's....really boring! It's just another iteration of "the thing we could already do, but in the browser." Jeff even says, "there are no VMs", and then has to clarify, "just WASM"(a VM). Then they share the screens and do the CRDT thing and he goes, "it's like we both SSH'd into a shared server, except there's no server!" But there IS a server - it's the one that's exposed by the service worker! Huh, yeah, it's almost as if we should go back and fix the UX around the actual OS instead of doing whatever this is.