Has unsafe_destructor landed? It's still needed.

[diwic]

According to rust-lang's latest blog post, the unsafe_destructor reform has landed, so I should not need to use unsafe_destructor. But still, with Rust nightly (2015-02-13 21:15:39) I still get an error for simple cases like this:

struct B<'a> {
   z: &'a str,
}

impl<'a> Drop for B<'a> {
   fn drop(&mut self) {
       println!("{}", self.z);
   }
}

[tomjakubowski]

The blog post was a little misleading; the RFC landed but not the changes to the compiler.

The specific issue for removing #[unsafe_destructor] is #22196

[steveklabnik]

This is what I thought, but wasn't sure. Thanks tom!

[pnkfelix]

To be precise (and also accurate):

• The analysis that supports the removal of #[unsafe_destructor] has landed in the compiler, and is applied to all destructors (including those that currently use the #[unsafe_destructor] attribute.
• ... but ...
• until we gain more confidence in the results, we are still requiring the use of the #[unsafe_destructor] attribute to implement Drop on generic types.

The is what #22196 is meant to convey, but I thought the phrase "but not the changes to the compiler" in @tomjakubowski's post above might be a little confusing to people, since the vast bulk of the necessary changes to the compiler have not only landed, but are also turned on!

[diwic]

Ok, that makes sense! Thanks for the clarification.