cc @SimonSapin

Looks good to me. Thanks!

It probably makes sense to implement it also for &'a mut, for completeness.

@Sawyer47 Don’t impls for &[T] also implicitly apply to &mut [T]?

It probably makes sense to implement it also for &'a mut, for completeness.

vec!(1, 2, 3).as_mut_slice().to_json() already works as it is. But perhaps we should also implement ToJson for &str?

@SimonSapin Honestly I'm not sure. There are separate implementations of Eq, Ord, PartialEq, PartialOrd, Show and Hash for both &'a and &'a mut so I thought the first doesn't imply the second.

Some traits are explicitly implemented for &mut T so that type itself can satisfy the trait. For example the following would not work with this patch:

&[&mut [1]].to_json()

Because &mut [T] doesn't implement ToJson, you can't satisfy any implementation of ToJson, so it fails to compile. It's probably not worth it to extend to &mut T for now.

@alexcrichton Thanks, good to know.