That is a very cool idea. Maybe a bit annoying to use, as you need a fixed size (I think?) special purpose allocator, and having these span across a page boundary doesn't seem possible. I definitely need to try this…
This code seems broken, though. The big challenge with these kinds of lock-free parallel algorithms is always how to deal with lifetimes and prevent use-after-frees (unless you never want to free anything from this…
That is a very cool idea. Maybe a bit annoying to use, as you need a fixed size (I think?) special purpose allocator, and having these span across a page boundary doesn't seem possible. I definitely need to try this…
This code seems broken, though. The big challenge with these kinds of lock-free parallel algorithms is always how to deal with lifetimes and prevent use-after-frees (unless you never want to free anything from this…