dcoughlin marked 5 inline comments as done.
================
Comment at: include/clang/StaticAnalyzer/Core/CheckerManager.h:96
@@ -95,1 +95,3 @@
+enum class ObjCCheckerKind {
+ PreVisit,
----------------
xazax.hun wrote:
> I do not really like the name ObjCCheckerKind, because it is not kind of an
> Obj-C related checker. It is the kind of the visit of the message expression.
> Maybe ObjCMessageVisitKind? Or just MessageVisitKind to be a bit shorter?
I've changed this to ObjCMessageVisitKind, as you suggested.
================
Comment at: lib/StaticAnalyzer/Core/ExprEngineObjC.cpp:153
@@ +152,3 @@
+ ProgramStateRef notNilState, nilState;
+ std::tie(notNilState, nilState) = State->assume(receiverVal);
+ if (nilState && !notNilState) {
----------------
xazax.hun wrote:
> The old code had a comment about merging two cases and a reference to a rdar.
> Is that rdar already fixed? Maybe it would be good to preserve the at least
> the first part of the commend?
Thanks for pointing this out! This revealed a deeper issue where the previous
patch did not drop the non-nil flow after a method call when it was unknown
whether the receiver was nil. I've fixed the behavior, added a comment
explaining why it is needed, and added a test for it.
http://reviews.llvm.org/D12123
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
