Range Redesign: Copy Semantics

[victoroak]

On Sunday, 21 January 2024 at 05:00:31 UTC, Jonathan M Davis 
wrote:
>
>  [...]
>
>

I think one problem with just "next" as the api for an Input 
Range is that currently I don't know any safe way to return a 
reference to the item. If "next" returns a pointer we can't be 
sure the backing data structure (a growable array like 
std::vector for example) did not freed that memory yet but I 
guess this is the same for any range right now.

Any way, it would be good to have an optimization for Nullable!T 
or a different type that when T is nullable (e.g. classes or 
pointers) it doesn't need a boolean flag but also loses the 
ability to store null inside it.

That's one of the optimizations of "Option" in Rust, although in 
Rust references can't be null so it's ok to do this.