This is the mail archive of the fortran@gcc.gnu.org mailing list for the GNU Fortran project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: First Patch


On 06/08/2011 07:16 PM, Tobias Burnus wrote:
First, please do *not* create a diff for the change log. The files tend
to change and then applying the diff is a bit extra work.

Ok. How do I give you the new Changelog then? Should I include it in my email? Or do I attach a new Changelog file?



Additionally,
you have seemingly added the changes to ./ChangeLog. However, for
gfortran one has to use ./gcc/fortran/ChangeLog - and for test cases
./gcc/testsuite/ChangeLog. The modifications in libgfortran/caf/* have
to be listed in ./libgfortran/ChangeLog.

Ok. So for this patch I'd be editing three different Changelog files:


./ligfortran/ChangeLog
./gcc/fortran/ChangeLog
./gcc/testsuite/ChangeLog



Ok, the attached file has all the changes you and the others have said
except for the test case. I took a look at the tests but I didn't
really understand how they work. Could you explain those to me? I'm
feeling a little stuck here.

The question is a bit broad, thus I am not sure how to answer.


Yeah, when you are feeling lost it is difficult to formulate a specific question. But your email was a good introduction, so I'm more clear how things work. I will read the wiki page you linked to next.


For run-time tests: Check the result via, e.g., "if (stat /= 0) call
abort()". If the assert is not true, one then gets an abort - and the
test fails.

Ok.


See also http://gcc.gnu.org/wiki/HowToPrepareATestcase

Ok, thanks.



However, I think it should be sufficient to copy coarray_2.f90 to
gfortran.dg/coarray/ - with the following modifications:
- Remove dg-options "-fcoarray=single". Reason: In ./coarray/, the
-fcoarray= flag is automatically added -- and also -fcoarray=lib
is/should be tested
- Remove the num_images() check (in case on has several images it would
fail); one might also remove the "ERROR STOP" line - in that case, also
the should-fail and the dg-output line have to be removed.


Ok. I've removed the "ERROR STOP", the dg-shouldfail and the dg-output.



You should also check that the tests in gfortran.dg/coarray/ work with
the MPI version, which is not automatically tested with "make
check-gfortran". The quickest way to do so under Linux is:

mpicc -c ../../../../libgfortran/caf/mpi.c
for I in *.f90; do echo ==== $I =====; mpif90 -fcoarray=lib $I mpi.o &&
mpiexec -np 3 ./a.out; done

Note: That assumes that "mpif90" calls the correct Fortran compiler; you
should check that "mpif90 -v" prints something like: gcc version 4.7.0
20110608 (correct version + date).


Hmm... all tests give the same error:

Fatal Error: Argument to -fcoarray is not valid: lib


Maybe there's a problem with my mpif90? This is what I have:


% mpif90 -v
Using built-in specs.
COLLECT_GCC=/usr/local/bin/gfortran
COLLECT_LTO_WRAPPER=/usr/local/gfortran/libexec/gcc/i686-pc-linux-gnu/4.6.0/lto-wrapper
Target: i686-pc-linux-gnu
Configured with: /home/jerry/gcc/trunk/configure --prefix=/usr/local/gfortran --enable-languages=c,c++,fortran --disable-libmudflap --enable-libgomp --disable-shared
Thread model: posix
gcc version 4.6.0 20110123 (experimental) [trunk revision 169143] (GCC)



Looks OK to me, but I don't know what I'm looking for.


Daniel.
--
I'm not overweight, I'm undertall.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]