James Liu

> @james7132 was the `SmallVec` in #74 problematic somehow? I'm interested in removing the `Vec` for small block counts and wondering if there's any particular direction we should explore next...

`FixedBitset::new` is now const so one of the two parts of this issue have been resolved.

I don't think it's possible to implement this as the `Index` implementation uses borrows on constants `&true` and `&false`. Providing a `&mut bool` return value is likely impossible.

`IndexMut` relies on `Index`'s associated type. We'd need to do the same for `Index` and I'm not sure if that's desirable from an API design perspective.

Not sure if this is a great idea due to the costs of monomorphization during compilation and the exposed generics as a part of it's type, which will need to...

@jrraymond I've updated this PR to only include the usize changes. Please take another look.

Thanks for the deep dive into this. What's the policy on this crate for the use of `unsafe`? We can convert a `&[usize]` into an equivalent `&[u32]` and then use...

Ultimately I think we should improve the benchmark coverage, and maybe convert them to Criterion for more stable results.

Attempted an implementation of explicit vectorization using a split like `glam` does. ``` group avx2 sse2 usize ----- ---- ---- ----- insert/1m 1.00 1772.7±8.57µs ? ?/sec 1.00 1777.2±13.75µs ? ?/sec...

Those tests seem to include the IO bottleneck of running `cargo clean`. Using https://doc.rust-lang.org/cargo/reference/timings.html might show some closer differences.