[PATCH] Delete various dead assignments

[Richard Smith]

================
Comment at: lib/AST/CommentCommandTraits.cpp:65
@@ -64,3 +64,2 @@
     BestCommand.push_back(Command);
-    BestEditDistance = EditDistance;
   }
----------------
Fixed in r203153.

================
Comment at: lib/Lex/ModuleMap.cpp:1965
@@ -1964,3 +1964,2 @@
       Diags.Report(StarLoc, diag::err_mmap_inferred_framework_submodule);
-      Framework = false;
     }
----------------
We don't seem to use the `Framework` flag in this function at all; this looks like a bug.

================
Comment at: lib/Sema/SemaDeclCXX.cpp:11461
@@ -11460,3 +11460,2 @@
         << TypeWithKeyword::getTagTypeKindName(Kind) << Name;
-      isExplicitSpecialization = true;
     }
----------------
This one looks fine to drop, but the code is arguably more robust to future changes with this dead store.

================
Comment at: lib/Sema/SemaTemplate.cpp:5932-5938
@@ -5931,9 +5931,6 @@
         << SourceRange(LAngleLoc, RAngleLoc);
-    else
-      isExplicitSpecialization = true;
   } else if (TUK != TUK_Friend) {
     Diag(KWLoc, diag::err_template_spec_needs_header)
       << FixItHint::CreateInsertion(KWLoc, "template<> ");
     TemplateKWLoc = KWLoc;
-    isExplicitSpecialization = true;
   }
----------------
It seems suspicious that `isExplicitSpecialization` is unused from this point onwards, but this needs more checking.

================
Comment at: tools/c-index-test/c-index-test.c:1332
@@ -1331,3 +1331,3 @@
       const char *RootParentName;
-      Root = Parent = p;
       do {
----------------
Obviously redundant, removed in r203169.

================
Comment at: utils/TableGen/NeonEmitter.cpp:765
@@ -764,3 +764,2 @@
     usgn = false;
-    poly = false;
     type = 'v';
----------------
I don't know if `poly` is supposed to be unused from this point onwards, but even if so, this dead store could make the code robust against future changes that add later uses of `poly`.


http://llvm-reviews.chandlerc.com/D2882