Per Vognsen on Nostr: While this has nice ergonomics, I take exception to the idea that it's zero-copy: ...
While this has nice ergonomics, I take exception to the idea that it's zero-copy: https://manishearth.github.io/blog/2022/08/03/zero-copy-3-so-zero-its-dot-dot-dot-negative/
As soon as your types have references (e.g. the Person struct example), the operating system loader has to do pointer fix-ups on load. And because this usually happens with the additional overhead of copy-on-write soft page faults, I could see this being significantly slower than equivalent non-copy-on-write pointer fix-up code. I'd have to do proper testing.
As soon as your types have references (e.g. the Person struct example), the operating system loader has to do pointer fix-ups on load. And because this usually happens with the additional overhead of copy-on-write soft page faults, I could see this being significantly slower than equivalent non-copy-on-write pointer fix-up code. I'd have to do proper testing.