rsmith added inline comments.
================
Comment at: lib/Sema/SemaDecl.cpp:10519-10520
@@ +10518,4 @@
+ auto *CE = dyn_cast<CXXConstructExpr>(Init);
+ bool DiagErr = (var->isInitKnownICE() || (CE &&
CE->getConstructor()->isConstexpr()))
+ ? !var->checkInitIsICE() : !checkConstInit();
+ if (DiagErr)
----------------
EricWF wrote:
> rsmith wrote:
> > In C++11 onwards, `checkInitIsICE` is the right thing to use in all cases.
> > In C++98, `checkInitIsICE` is appropriate for globals of integral type; we
> > do not have an implementation of strict constant expression checking for
> > other types. Perhaps just documenting that (and maybe adding a FIXME here)
> > is the best you can do for now, assuming you're not interested in
> > implementing the C++98 rules.
> SGTM. I think I'll fall back to `isConstantInitializer()` in C++03. Do you
> have an example of what `checkConstantInitializer()` gets wrong in C++03?
struct A {
~A();
int n;
} a;
int *p = &a.n; // constant initializer, but not a C++98 address constant
expression
https://reviews.llvm.org/D23385
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
