Unadorned path should resolve same as relative path written w.r.t. self/super

[alexcrichton]

The following test should have all imports as valid:

use a;        // bad
use self::a;  // good
use a::b;     // good

mod a {
  pub type b = int;
}
mod b {
  use a;        // bad
  use super::a; // good
  use a::b;     // good
}

Apparently resolve isn't realizing that use a is equivalent to use super::a. A little more discussion can be found in #6226

[pnkfelix]

Nominating for milestone, "Well defined". (In particular, I cannot tell if this is a true bug or some sort of weird expected behavior without a more explicit definition of the semantics of path resolution.)

[pnkfelix]

See also #7551 (which I just filed and arguably may be more appropriate for milestone 1 than this bug actually is).

[pnkfelix]

Visiting for triage, email 2013-08-12. nothing more to add.

[pnkfelix]

Note also #4293

[catamorphism]

Metabug for resolve surprises: #6143

[nikomatsakis]

This is not the expected behavior afaik

[catamorphism]

De-milestoning this since metabug is on maturity 1

[alexcrichton]

Closing as a dupe of #8215, I'm not entirely sure that all imports should be valid in that program, I will re-post with an updated version in 8215