[Patch] Fix for PR18746 -rejects-valid on a constexpr member function of a class template with a deduced return type and no return statements
Rahul Jain
1989.rahuljain at gmail.com
Thu Apr 10 18:29:24 PDT 2014
Gentle ping.
On Apr 8, 2014 11:26 PM, "Rahul Jain" <1989.rahuljain at gmail.com> wrote:
>
> Hi Richard,
>
> Hi all,
>
> This patch fixes PR18746.
>
>
> Index: lib/Sema/SemaDeclCXX.cpp
> ===================================================================
> --- lib/Sema/SemaDeclCXX.cpp (revision 205775)
> +++ lib/Sema/SemaDeclCXX.cpp (working copy)
> @@ -1177,12 +1177,19 @@
> // C++1y doesn't require constexpr functions to contain a 'return'
> // statement. We still do, unless the return type is void, because
> // otherwise if there's no return statement, the function cannot
> - // be used in a core constant expression.
> + // be used in a core constant expression. Also, deduced return type
> + // with no return statements are perfectly legal. For example:
> + // template<typename T> struct X { constexpr auto f() {} };
> +
> bool OK = getLangOpts().CPlusPlus1y &&
> Dcl->getReturnType()->isVoidType();
> +
> + if (!getLangOpts().CPlusPlus1y ||
> + !Dcl->getReturnType()->getContainedAutoType()) {
> Diag(Dcl->getLocation(),
> OK ? diag::warn_cxx11_compat_constexpr_body_no_return
> : diag::err_constexpr_body_no_return);
> return OK;
> + }
> }
> if (ReturnStmts.size() > 1) {
> Diag(ReturnStmts.back(),
> Index: test/SemaCXX/cxx1y-deduced-return-type.cpp
> ===================================================================
> --- test/SemaCXX/cxx1y-deduced-return-type.cpp (revision 205775)
> +++ test/SemaCXX/cxx1y-deduced-return-type.cpp (working copy)
> @@ -261,6 +261,7 @@
>
> namespace Constexpr {
> constexpr auto f1(int n) { return n; }
> + template<typename T> struct X { constexpr auto f() {} }; // PR18746
> struct NonLiteral { ~NonLiteral(); } nl; // expected-note
> {{user-provided destructor}}
> constexpr auto f2(int n) { return nl; } // expected-error {{return type
> 'Constexpr::NonLiteral' is not a literal type}}
> }
>
>
> Please if someone could review the same.
>
> Thanks,
> Rahul
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20140411/1ab12145/attachment.html>
More information about the cfe-commits
mailing list