Re: [patch,fortran] Allow namelist with internal file (PR 28224)

[Tobias Burnus <burnus at net-b dot de>]

Hi,

Paul Thomas wrote:
>> - How to run only the gfortran test suit? I found check-gcc, but no
>> check-(g)fortran.   
> make check-gfortran
This does not work for me. I do a:
./configure --enable-languages=c,fortran
make
make install

and in the same directory "make check". In this directory also
check-gcc, check-libcpp, check-gcc-c++, check-c++,
check-target-libgfortran work. (make check-target-libgfortran does not
do any test suit runs, though.)
But I get: make check-gfortran
make: *** No rule to make target `check-gfortran'.  Stop.

>> - I get   "XPASS: gfortran.dg/csqrt_2.f (PR24313)."
>>   Is it time to remove the "{ xfail *-*-linux-gnu }"?
>>   That it passes I also see when looking at the results of Andreas
>> Jaeger,
>>   but, well, he also uses an openSUSE system with glibc 3.5.   
> This is seen on newer distros - something somewhere, I do not recall
> where, has been fixed.
According to http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24313
this was a bug in glibc 2.3.5, which got fixed 2005-10-13, see
http://sources.redhat.com/bugzilla/show_bug.cgi?id=1466
The PR also links to a cacosh bug, which was fixed 2006-08-03
http://sources.redhat.com/bugzilla/show_bug.cgi?id=2182

If I understand XPASS and {xfail}  correctly, one should remove the
{xfail} if the bug has been fixed.

Compare: http://gcc.gnu.org/wiki/HowToPrepareATestcase
"Each testcase which is XFAIL-ed must have a corresponding PR open in
the bug database, and the PR number must be referenced in aEach testcase
which is XFAIL-ed must have a corresponding PR open in the bug database,
and the PR number must be referenced in a comment in the testcase
itself. This makes sure that the failure is not forgotten. comment in
the testcase itself. This makes sure that the failure is not forgotten."

Our bug PR is closed as Resolved->Invalid. Should we now remove the
{xfail} ?

> I'll take a look at your patch this afternoon.
Thanks.

Tobias