Message ID | 20181204084251.GF12380@tucnak |
---|---|
State | New |
Headers | show |
Series | Fix P0634R3 - Down with typename! testsuite fallout | expand |
On Tue, Dec 04, 2018 at 09:42:51AM +0100, Jakub Jelinek wrote: > Hi! > > I've noticed these two tests now FAIL with -std=c++2a, they don't print > any diagnostics. As I believe that was the point of the P0634R3 changes, > this patch adjusts the testcases. Tested on x86_64-linux, ok for trunk? > > 2018-12-04 Jakub Jelinek <jakub@redhat.com> > > * g++.old-deja/g++.oliva/typename1.C: Don't expect any diagnostics > for C++2a. > * g++.old-deja/g++.oliva/typename2.C: Likewise. The patch is fine, sorry for missing that. Apparently I only ran c++2a with dg.exp. Marek
--- gcc/testsuite/g++.old-deja/g++.oliva/typename1.C.jj 2008-09-05 12:54:51.000000000 +0200 +++ gcc/testsuite/g++.old-deja/g++.oliva/typename1.C 2018-12-04 09:36:46.137087008 +0100 @@ -12,5 +12,5 @@ template <class T> struct bar { }; template <class T> struct baz { - typedef bar<T>::foo foo; // { dg-error "" } missing typename + typedef bar<T>::foo foo; // { dg-error "need 'typename' before" "" { target c++17_down } } }; --- gcc/testsuite/g++.old-deja/g++.oliva/typename2.C.jj 2008-09-05 12:54:51.000000000 +0200 +++ gcc/testsuite/g++.old-deja/g++.oliva/typename2.C 2018-12-04 09:37:27.069419970 +0100 @@ -23,6 +23,6 @@ template <class T> struct bar { }; template <class T> struct baz { - typedef bar<T>::foo foo; // { dg-error "" } implicit typename + typedef bar<T>::foo foo; // { dg-error "need 'typename' before" "" { target c++17_down } } void m(foo); };