Alex, you are right. I need to rephrase my question. Consider the following:
pub trait T1 {}
pub trait T2 {}
impl T1 for ~T2 {}
pub struct S;
impl T2 for S {}
fn check_bounds<U: T1>(_: U) {}
fn main() {
let x = ~S as ~T2:Send;
check_bounds(x);
}
Should check_bounds fail? It currently compiles, but according to
https://github.com/mozilla/rust/issues/5781, trait matching should be
invariant. If that's true, ~T2 and ~T2:Send would be different so here we
only have a T1 implementation for ~T2, not ~T2:Send, an error.
Then task::try(...).unwrap() should fail, too, if my interpretation is
correct.
On Fri, Apr 18, 2014 at 12:10 AM, Alex Crichton <a...@crichton.co> wrote:
> The ~Any type has a special implementation of Show:
>
> https://github.com/mozilla/rust/blob/master/src/libstd/any.rs#L151-L155
>
> I believe it was primarily used in failure messages originally (you
> can fail a task with ~Any)
>
> On Thu, Apr 17, 2014 at 8:55 AM, Edward Wang <edward.yu.w...@gmail.com>
> wrote:
> > It current can compile, but judging from signatures:
> >
> > std::task::try is pub fn try<T: Send>(f: proc(): Send -> T) -> Result<T,
> > ~Any:Send>
> > std::result::unwrap on the other hand is impl<T, E: Show> Result<T, E> {
> fn
> > unwrap(self) -> T {...} }
> >
> > There's no way the error part of result from task::try(...) can fulfil
> Show
> > so it shouldn't compile.
> >
> > Though to ask the list first before filing a bug report.
> >
> > Regards,
> > Edward
> >
> > _______________________________________________
> > Rust-dev mailing list
> > Rust-dev@mozilla.org
> > https://mail.mozilla.org/listinfo/rust-dev
> >
>
_______________________________________________
Rust-dev mailing list
Rust-dev@mozilla.org
https://mail.mozilla.org/listinfo/rust-dev
