rjmccall added inline comments.
================
Comment at: lib/Sema/SemaPseudoObject.cpp:57
@@ +56,3 @@
+ // with a base limits the number of cases here.
+ assert(refExpr->isObjectReceiver());
+
----------------
Well, it should be okay to call this on something that doesn't have a base; it
just doesn't need to do anything and can just return refExpr.
================
Comment at: lib/Sema/SemaPseudoObject.cpp:1591
@@ +1590,3 @@
+ return baseOVE->getSourceExpr();
+ } else if (ObjCSubscriptRefExpr *refExpr =
+ dyn_cast<ObjCSubscriptRefExpr>(opaqueRef)) {
----------------
That's what I'm saying, though: the capture process should be building the
syntactic MSPropertySubscriptExpr using the captured OVEs for the index
expressions as well as the recursively-captured base, and Rebuilder will then
need to undo that.
We want to consistently replace all the captured sub-trees with their OVEs so
that clients can easily recover the relationship between the syntactic and
semantic forms.
================
Comment at: lib/Sema/SemaPseudoObject.cpp:1621
@@ -1582,1 +1620,3 @@
+ };
+ return Rebuilder(S, Callback).rebuild(E);
}
----------------
My hope is that you'll be able to get this whole function down to just this
last line and won't need to split out any of the cases here.
http://reviews.llvm.org/D13336
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
