resolve: Try to fix instability in import suggestions

[petrochenkov]

cc #42033

lookup_import_candidates walks module graph in DFS order and skips modules that were already visited (which is correct because there can be cycles).
However it means that if we visited std::prelude::v1::Result::Ok first, we will never visit std::result::Result::Ok because Result will be skipped as already visited (note: enums are also modules here), and otherwise, if we visited std::result::Result::Ok first, we will never get to std::prelude::v1::Result::Ok.
What child module of std (prelude or result) we will visit first, depends on randomized hashing, so we have instability in diagnostics.

With this patch modules' children are visited in stable order in lookup_import_candidates, this should fix the issue, but let's see what Travis will say.

r? @oli-obk

[oli-obk]

Even if this becomes deterministic by these changes, I dislike the results:

+ = help: there is an enum variant `std::result::Result::Ok`, try using `std::result::Result`?
+ = help: there is an enum variant `std::prelude::v1::Option::Some`, try using `std::prelude::v1::Option`?

Result shows up in the real path while Option shows up in the reexport in the prelude.

I still think that it's impossible to produce nice diagnostics with paths during resolve and that we should be caching them until resolve is finished and report them then

[petrochenkov]

I don't see how delaying will help here.
These particular diagnostics are reported during the second stage of resolve when all import resolution is done and all definitions are available. Fully completing resolution will not add any new information useful here.

The problem with prelude paths is that reexports are shown in candidates as well as true definitions. The intent of the original PR was to filter out the reexports, however the implementation doesn't work correctly and nothing outside of the current crate is considered a reexport and filtered out.
fn each_child_of_item and friends should preserve the export/non-export distinction and provide it to resolve so it can filter away reexports properly.
This is an orthogonal issue though and it doesn't block migration of compile-fail tests to UI testing.

[oli-obk]

Makes sense. Implementation Lgtm.

You'll need a reviewer to sign off the PR though.

[petrochenkov]

r? @jseyfried

[jseyfried]

@bors r+

[carols10cents]

@bors r=jseyfried

[bors]

📌 Commit a6993d6 has been approved by jseyfried

[bors]

⌛ Testing commit a6993d6 with merge ca764ee39451f96aa9db27b392b6d1b1974676c3...

[bors]

💔 Test failed - status-travis

[alexcrichton]

@bors: retry

• musl segfault Spurious segfault in tests for musl #38618

[bors]

⌛ Testing commit a6993d6 with merge 6e8452e...

[bors]

☀️ Test successful - status-appveyor, status-travis
Approved by: jseyfried
Pushing 6e8452e to master...