Merge #693

693: Name resolution refactoring r=matklad a=flodiebold

This is still very WIP, but it's becoming quite big and I want to make sure this isn't going in a completely bad direction :sweat_smile:. I'm not really happy with how the path resolution looks, and I'm not sure `PerNs<Resolution>` is the best return type -- there are 'this cannot happen in the (types/values) namespace' cases everywhere. I also want to unify the `resolver` and `nameres` namespaces once I'm done switching everything to `Resolver`. Also, `Resolver` only has a lifetime because it needs to have a reference to the `ItemMap` during import resolution :confused: 

The differences in the completion snapshots are almost completely just ordering (except it completes `Self` as well now), so I changed it to sort the completions before snapshotting.

Co-authored-by: Florian Diebold <[email protected]>

1 files changed, 4 insertions, 4 deletions

diff --git a/crates/ra_hir/src/ty/snapshots/tests__infer_function_generics.snap b/crates/ra_hir/src/ty/snapshots/tests__infer_function_generics.snap
index 8ff6e55a6..91c48897c 100644
--- a/crates/ra_hir/src/ty/snapshots/tests__infer_function_generics.snap
+++ b/crates/ra_hir/src/ty/snapshots/tests__infer_function_generics.snap
@@ -1,12 +1,12 @@
 ---
-created: "2019-01-26T18:16:16.530712344+00:00"
+created: "2019-01-27T14:52:29.934503829+00:00"
 creator: [email protected]
 expression: "&result"
 source: crates/ra_hir/src/ty/tests.rs
 ---
-[10; 11) 't': [unknown]
-[21; 26) '{ t }': [unknown]
-[23; 24) 't': [unknown]
+[10; 11) 't': T
+[21; 26) '{ t }': T
+[23; 24) 't': T
 [38; 98) '{     ...(1); }': ()
 [44; 46) 'id': fn id<u32>(T) -> T
 [44; 52) 'id(1u32)': u32