Can't use #[deriving(Clone)] on struct with Clone bounds

[japaric]

STR

#[deriving(Clone)]
//~^ error: trait `Clone` already appears in the list of bounds [E0127]
struct Foo<'a, T: 'a + Clone> {
    foo: &'a [T],
}

fn main() {}

Output

If you look at the --pretty=expanded version, you'll see T is bounded twice by the Clone trait

#![feature(phase)]
#![no_std]
#![feature(globs)]
#[phase(plugin, link)]
extern crate "std" as std;
#[prelude_import]
use std::prelude::*;
//~^ error: trait `Clone` already appears in the list of bounds [E0127]
struct Foo<'a, T: 'a + Clone> {
    foo: &'a [T],
}
#[automatically_derived]
impl <'a, T: ::std::clone::Clone + 'a + Clone> ::std::clone::Clone for
     Foo<'a, T> {
    #[inline]
    fn clone(&self) -> Foo<'a, T> {
        match *self {
            Foo { foo: ref __self_0_0 } =>
            Foo{foo: ::std::clone::Clone::clone(&(*__self_0_0)),},
        }
    }
}

Version

rustc 0.13.0-dev (82fc1aa87 2014-11-27 10:11:19 +0000)

The obvious fix seems to make the deriving syntax extension filter out the duplicated Clone just by looking at the trait name. The problem is that ::std::clone::Clone may not be the same trait as Clone, but AFAIK the syntax extension can't know that, since that (path) information is collected after macro expansion.

Work-around

One workaround to this issue is using the fact that the deriving syntax extension doesn't look at where bounds (see #19358), the following code compiles:

#[deriving(Clone)]
struct Foo<'a, T: 'a> where T: Clone {
    foo: &'a [T],
}

fn main() {}

[alexcrichton]

This now works, so I think it was fixed in the meantime? Yay!