test case for closed bug now causes ICE

[jbclements]

The test case for issue #3979 was xfailed (and didn't parse), but that issue is closed, so I fixed up the syntax and ran it... and now it causes an ICE. I've committed the updated test, but it may be a while before that winds up in incoming, so I'll paste the text of the test here too:

// xfail-test
trait Positioned<S> {
  fn SetX(&mut self, S);
  fn X(&self) -> S;
}

#[allow(default_methods)]
trait Movable<S, T>: Positioned<T> {
  fn translate(&self, dx: T) {
    self.SetX(self.X() + dx);
  }
}

struct Point { x: int, y: int }

impl Positioned<int> for Point {
    fn SetX(&mut self, x: int) {
        self.x = x;
    }
    fn X(&self) -> int {
        self.x
    }
}

impl Movable<int, int> for Point;

pub fn main() {
    let p = Point{ x: 1, y: 2};
    p.translate(3);
    assert!(p.X() == 4);
}

[metajack]

@jbclements Is this sufficiently serious to warrant nominating?

[emberian]

Still a bug.

[emberian]

(@catamorphism)

[jbclements]

Oog... I should have recorded the ICE. Now I get this compilation error:

Running /usr/local/bin/rustc:
/tmp/ggg.rs:9:14: 9:27 error: binary operation + cannot be applied to type `U`
/tmp/ggg.rs:9     self.SetX(self.X() + dx);
                            ^~~~~~~~~~~~~
error: aborting due to previous error

... which might be because the ICE was fixed, or might be because something about the language changed to cause an earlier breakage. Sigh.

[msullivan]

ICE isn't fixed.

[msullivan]

OK, I think your latest failure is due to language changes, and the test can be updated. I think that the test case was xfailed because it was broken, due to #4396. I think that I fixed the reason that it was originally xfailed with previous default method work (although there still generics problems with supertraits that I am working on at the moment).

Dropping the &muts and getting rid of the assignment in SetX makes the test pass. issue-3979-xcrate.rs has the same failure. issue-3979.rs, on the other hand, works fine, although it was xfailed with a comment saying it was because of "ICE with explicit self". That was added by @pcwalton while demuting the test suite, in 82062a6.

[msullivan]

Here is an updated version of issue-3979-generics.rs:

// Copyright 2012 The Rust Project Developers. See the COPYRIGHT
// file at the top-level directory of this distribution and at
// http://rust-lang.org/COPYRIGHT.
//
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
// http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
// <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
// option. This file may not be copied, modified, or distributed
// except according to those terms.

trait Positioned<S> {
  fn SetX(&mut self, S);
  fn X(&self) -> S;
}

trait Movable<S: Add<S, S>>: Positioned<S> {
  fn translate(&mut self, dx: S) {
    self.SetX(self.X() + dx);
  }
}

struct Point { x: int, y: int }

impl Positioned<int> for Point {
    fn SetX(&mut self, x: int) {
        self.x = x;
    }
    fn X(&self) -> int {
        self.x
    }
}

impl Movable<int> for Point;

pub fn main() {
    let p = Point{ x: 1, y: 2};
    p.translate(3);
    assert_eq!(p.X(), 4);
}

[msullivan]

OK, got it. The ICE is occuring while trying to report a borrow check error in translate. The self.X() + dx needs to be hoisted out of the method call.

I am going to fix the bugs in these tests, close this issue, and open a new one with a reduced test case for the real problem.