I dislike this - having the “array” be a struct containing the pointer and size fields makes it easy to “copy” the array such that you get dangling pointers. Si

I dislike this - having the “array” be a struct containing the pointer and size fields makes it easy to “copy” the array such that you get dangling pointers. Similarly there’s no way to track lifetime or ownership of the array. There are long term ABI benefits to these data structures just being opaque pointers outside of the implementing libraries