On Tue, Apr 16, 2019 at 02:51:14PM +0200, Jan Hubicka wrote:
> > Hi Jan,
> >
> > The test causes
> >
> > WARNING: lto.exp does not support dg-do
> > WARNING: lto.exp does not support dg-options in primary source file
> >
> > This is fixed by the following patch
> >
> > --- ../_clean/gcc/testsuite/g++.dg/lto/pr89358_0.C 2019-04-15
> > 00:04:48.000000000 +0200
> > +++ gcc/testsuite/g++.dg/lto/pr89358_0.C 2019-04-16 13:13:14.000000000
> > +0200
> > @@ -1,5 +1,5 @@
> > -/* { dg-do link } */
> > -/* { dg-options "-std=c++17" } */
> > +/* { dg-lto-do link } */
> > +/* { dg-lto-options "-std=c++17" } */
>
> Thanks, this changle is OK.
> Honza
> > #include <map>
> >
> > extern void test();
> > --- ../_clean/gcc/testsuite/g++.dg/lto/pr89358_1.C 2019-04-15
> > 00:04:48.000000000 +0200
> > +++ gcc/testsuite/g++.dg/lto/pr89358_1.C 2019-04-16 13:14:44.000000000
> > +0200
> > @@ -1,4 +1,3 @@
> > -/* { dg-options "-std=c++14" } */
But this change is not ok and makes the test useless.
The whole point of the test is to test if compiling one TU with -std=c++17
and another one with -std=c++14 and linking them together doesn't produce
warnings. I think dg-options in *_1.C is ok, several tests have those:
20100302_1.C:// { dg-options "-fabi-version=0" }
pr60150_1.C:// { dg-options "-fno-lto" }
pr63270_2.C:// { dg-options "-fno-lto" }
pr64076_1.C:// { dg-options -fno-lto }
pr65276_1.C:// { dg-options "-O2" }
pr68811_1.C:// { dg-options "-O2 -flto -w" }
Jakub
