Bounds-checking for Fortran testsuite was: Re: [gfortran, testsuite] Fix bounds checking for substrings, enable bounds chekcing in dg testsuite, assorted fixes

[Janis Johnson]

On Fri, Dec 15, 2006 at 03:35:07PM -0800, Janis Johnson wrote:
> On Thu, Dec 14, 2006 at 05:09:49PM -0800, Janis Johnson wrote:
> > On Fri, Dec 15, 2006 at 12:29:28AM +0100, Andreas Tobler wrote:
> > > Tobias Schlüter wrote:
> > > >Andreas Tobler <toa@pop.agri.ch> wrote on Thu, 14 Dec 2006:
> > > >>+load_lib gcc-dg.exp
> > > >> load_lib fortran-torture.exp
> > > >>-load_lib gcc-dg.exp
> > > >
> > > >Unfortunately, this will make gcc-dg's TORTURE_OPTIONS override
> > > >fortran-torture's.
> > >
> > > I did not say it's correct, I only stated, it helped here to continue
> > > testing. I continue to investigate.
> >
> > This is OK for now to get tests running again.
> 
> Never mind, I checked in a different fix; see
> http://gcc.gnu.org/ml/gcc-patches/2006-12/msg01141.html.

That message mentioned a problem with TORTURE_OPTIONS.

Most uses of TORTURE_OPTIONS throughout the testsuite use the list
defined in lib/gcc-dg.exp because that's the file that is loaded
first.  For the Fortran tests it's loaded for the gfortran.dg tests,
and the variable keeps that definition for gfortran.fortran-torture.
If the torture tests are run alone (via RUNTESTFLAGS=whatever.exp)
then they get the list defined in fortran-torture.exp.

This patch fixes the Fortran tests to use the list defined in
fortran-torture.exp, but it's not yet fully tested.  I'll test it
next week unless someone else beats me to it.

Janis

Index: gcc/testsuite/gfortran.dg/dg.exp
===================================================================
--- gcc/testsuite/gfortran.dg/dg.exp	(revision 119939)
+++ gcc/testsuite/gfortran.dg/dg.exp	(working copy)
@@ -17,6 +17,7 @@
 # GCC testsuite that uses the `dg.exp' driver.
 
 # Load support procs.
+load_lib fortran-torture.exp
 load_lib gfortran-dg.exp
 
 # If a testcase doesn't have special options, use these.