Add name bindings for bad imports

[dirk]

WIP implementation of #31209.

The goal is to insert fake/dummy definitions for names that we failed to import so that later resolver stages won't complain about them.

[rust-highfive]

r? @alexcrichton

(rust_highfive has picked a reviewer for you, use r? to override)

[alexcrichton]

r? @nrc

I have a feeling you'll have a much better idea what's going on here than I

[nrc]

lgtm, what is still left to do? Adding a test or two seems like the only obvious thing to me.

[dirk]

@nrc: I'll definitely add some tests! 😉 Overall, though, do you think I took the right approach with adding the additional fields to ImportResolvingError and injecting into the module at error-resolution time?

[nrc]

Yeah, I think it is a fine approach. I'm slightly surprised there isn't an error def in there somewhere. In particular if a name ends up mapping to one of these dummy imports, how do we prevent type checking issuing further errors about it?

[dirk]

I'm slightly surprised there isn't an error def in there somewhere.

I was surprised about that myself. It may turn out we need to make one. I'll write up tests for this tonight and we can see how those turn out.

In particular if a name ends up mapping to one of these dummy imports, how do we prevent type checking issuing further errors about it?

I'm still getting familiar with the compiler internals so this may be completely wrong, but doesn't the compiler exit after reporting all the name/import/etc. resolution errors and not progress on to the type-checking stage?

[nrc]

doesn't the compiler exit after reporting all the name/import/etc. resolution errors

It used to, but I've been working on this recently and now will get all the way to type checking before bailing out most of the time.

[dirk]

@nrc: As of 250ffcb it actually works! 🎉

$ cat test.rs
use foo::bar;

fn main() {
    bar();
}

$ x86_64-apple-darwin/stage1/bin/rustc test.rs
test.rs:1:5: 1:8 error: unresolved import `foo::bar`. Maybe a missing `extern crate foo`? [E0432]
test.rs:1 use foo::bar;
              ^~~
test.rs:1:5: 1:8 help: run `rustc --explain E0432` to see a detailed explanation
error: aborting due to previous error

Now just to write some proper test(s) for it.

[dirk]

@nrc: Tests are updated for the new behavior; I think this is ready for your review.

[nrc]

nit: the previous style is more idiomatic

[dirk]

@nrc: Fixed! 😄

[nrc]

s/resolve_import_resolving_error/import_resolving_error

[nrc]

LGTM. Could you fix the remaining nits and squash the commits please?

[dirk]

@nrc: Fixed and squashed!

[nrc]

Thanks!

@bors: r+

[bors]

📌 Commit 026bcbf has been approved by nrc

[bors]

⌛ Testing commit 026bcbf with merge 8c77ffb...

[bors]

☀️ Test successful - auto-linux-32-nopt-t, auto-linux-32-opt, auto-linux-64-debug-opt, auto-linux-64-nopt-t, auto-linux-64-opt, auto-linux-64-x-android-t, auto-linux-cross-opt, auto-linux-musl-64-opt, auto-mac-32-opt, auto-mac-64-nopt-t, auto-mac-64-opt, auto-mac-ios-opt, auto-win-gnu-32-nopt-t, auto-win-gnu-32-opt, auto-win-gnu-64-nopt-t, auto-win-gnu-64-opt, auto-win-msvc-32-opt, auto-win-msvc-64-opt