Hi Juan,

I have faced similar challenges when calling Fortran code from C++ templates.

In the solution I came up with, I use a marked up Fortran template code and Cmake's configure_file command to generate a module for each combination of C++ types I need. The markup uses a token for each template argument and a decorator that is appended to Fortran symbols and is used in the module and file names to differentiate all of the CMake generated instantiations.

here is the relevant Cmake code that does all the work, (note: I have two template arguments one for coordinates and variables):

   set(teca_alg_f90_srcs)

   set(teca_alg_f90_generics
        gfdl_spline
        gfdl_tc_candidates
        )

   set(f_type real)
   set(c_types float double)
   foreach(generic_src ${teca_alg_f90_generics})
        foreach(c_type_var ${c_types})
            set(iso_c_type_var "${f_type}(c_${c_type_var})")
            string(SUBSTRING ${c_type_var} 0 1 var_name)
            foreach(c_type_coord ${c_types})
                string(SUBSTRING ${c_type_coord} 0 1 coord_name)
                set(decorator "c${coord_name}_v${var_name}")
                set(iso_c_type_coord "${f_type}(c_${c_type_coord})")
                configure_file(${generic_src}.f90.in
   ${generic_src}_${decorator}.f90 @ONLY)
                list(APPEND teca_alg_f90_srcs
   ${generic_src}_${decorator}.f90)
            endforeach()
        endforeach()
   endforeach()

   add_library(teca_alg ${teca_alg_cxx_srcs} ${teca_alg_f90_srcs})

And here is a link to an example marked up Fortran template code.

   https://github.com/LBL-EESA/TECA/blob/master/alg/gfdl_tc_candidates.f90.in

From C++ I use macros to declare decorated Fortran functions and define C++ overloads that call them.

   https://github.com/LBL-EESA/TECA/blob/master/alg/gfdl_tc_candidates.h

It's a bit of a heavy handed solution, but I think you can solve your problem in a similar way. I'm interested in hearing of other approaches that have worked.

Burlen


On 06/26/2017 01:40 PM, Juan E. Sanchez wrote:
Hi,

It seems like cmake cannot handle the case where the module name is the result of a macro. I am using this approach to compile the same code for different floating point precision. Any advice appreciated. This approach would apply to hundreds of files.

The error is:
Error copying Fortran module "concat". Tried "CONCAT.mod" and "concat.mod".
make[2]: *** [CMakeFiles/flib.dir/baz.o.provides.build] Error 1
make[1]: *** [CMakeFiles/flib.dir/all] Error 2
make: *** [all] Error 2


Regards,

Juan


baz.F:
#include "fmacros.inc"

      module CONCAT(baz)
      contains
      subroutine car(t3)
      end subroutine
      end module CONCAT(baz)

foo.F:
#include "fmacros.inc"

      module CONCAT(foo)
      contains
      subroutine bar(t1, t2)
      use CONCAT(baz)
      implicit none
      REAL(kind=8) t1
      REAL(kind=DWIDTH) t2
      call baz(t2)
      end subroutine
      end module CONCAT(foo)

fmacros.inc:

#define CONCAT(a) a/**/_double
#define DWIDTH 8

main.cc:
extern "C" {
void
}

CMakeLists.txt:

PROJECT(foo)
ENABLE_LANGUAGE(CXX Fortran)

ADD_LIBRARY(flib foo.F baz.F)




-- 

Powered by www.kitware.com

Please keep messages on-topic and check the CMake FAQ at: 
http://www.cmake.org/Wiki/CMake_FAQ

Kitware offers various services to support the CMake community. For more 
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/cmake

Reply via email to