[Trilinos-Users] Compiling on a blue gene

[Perschbacher, Brent M]

I think Brad's suggestion is about as good as I can offer too for tracking this down. The only other thing I am curious about is what is the path to the xlf95_r compiler that cmake found? It seems highly unlikely, but since you didn't give an exact path it is possible that cmake found a different version for xlf95_r than it did for xlC_r. I expect that isn't the problem, but I'd feel really silly if this was the cause and I didn't ask :)

Brent


On 11/11/09 7:56 AM, "Brad King" <brad.king at kitware.com> wrote:

John R. Cary wrote:
> cmake -D CMAKE_Fortran_COMPILER:FILEPATH=xlf95_r -D
> CMAKE_CXX_COMPILER:FILEPATH=xlC_r -D CMAKE_C_COMPILER:FILEPATH=xlc_r
[snip]
> /gpfs/software/linux-sles10-ppc64/apps/ibmcmp-aug2009/vacpp/bg/9.0/bin/xlC_r
> CMakeFiles/VerifyFortranC.dir/main.c.o
> CMakeFiles/VerifyFortranC.dir/VerifyC.c.o
> CMakeFiles/VerifyFortranC.dir/VerifyFortran.f.o
> CMakeFiles/VerifyFortranC.dir/VerifyCXX.cxx.o  -o VerifyFortranC
> -L/soft/apps/ibmcmp-aug2009/xlf/bg/11.1/lib -lgcc_eh -lgcc_eh -lxlf90
> -lxlomp_ser -lxlfmath -ldl -lrt -lgcc_eh
>     CMakeFiles/VerifyFortranC.dir/main.c.o: In function `main':
>
> /gpfs/home/projects/facets/intrepid/contrib/cmake-2.8.0-rc3-ser/share/cmake-2.8/Modules/FortranCInterface/Verify/main.c:(.text+0x14):
> undefined reference to `VerifyFortran'

> Looks like a Fortran name mangling issue.

Look at CMakeFiles/FortranCInterface/Output.cmake to see what mangling
was detected, if any.  The XL compiler's mangling is supposed to be
known.  What is the output of this:

  nm CMakeFiles/FortranCInterface/libmyfort.a

?

Thanks,
-Brad


_______________________________________________
Trilinos-Users mailing list
Trilinos-Users at software.sandia.gov
http://software.sandia.gov/mailman/listinfo/trilinos-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://software.sandia.gov/pipermail/trilinos-users/attachments/20091111/6c678093/attachment.html