Linking external C code to Octave is relatively simple, as the C functions can easily be called directly from C++. One possible issue is that the declarations of the external C functions may need to be explicitly defined as C functions to the compiler. If the declarations of the external C functions are in the header foo.h, then the tactic to ensure that the C++ compiler treats these declarations as C code is
#ifdef __cplusplus extern "C" { #endif #include "foo.h" #ifdef __cplusplus } /* end extern "C" */ #endif
When calling functions that are implemented in Fortran code, some pecularities
have to be taken into account. Symbol names in Fortran are case-insensitive,
and depending on the used Fortran compiler, function names are either exported
with all lower-case or with all upper-case characters. Additionally, some
compilers append none, one or two underscores "_
" at the end of
exported function names. This is called "name-mangling".
Octave supplies macros that allow writing code that automatically handles the
name-mangling for a number of different Fortran compilers. These macros are
F77_FUNC
and F77_FUNC_
. The former should be used for Fortran
functions that do not contain any underscores in their name. The latter should
be used for Fortran functions with underscores in their names. Both macros
take two arguments: The first is the Fortran function name in all lower-case
characters. The second is the same Fortran function name in all upper-case
characters.
Additionally to the name-mangling, different compilers are using different calling conventions for some types. Octave defines the following preprocessor macros to allow writing code that can be used with different Fortran calling conventions.
Note that we don’t attempt to handle Fortran functions, we always use subroutine wrappers for them and pass the return value as an extra argument.
Use the following macros to pass character strings from C to Fortran:
F77_CHAR_ARG(x) F77_CONST_CHAR_ARG(x) F77_CXX_STRING_ARG(x) F77_CHAR_ARG_LEN(l) F77_CHAR_ARG_DECL F77_CONST_CHAR_ARG_DECL F77_CHAR_ARG_LEN_DECL
Use the following macros to write C-language functions that accept Fortran-style character strings:
F77_CHAR_ARG_DEF(s, len) F77_CONST_CHAR_ARG_DEF(s, len) F77_CHAR_ARG_LEN_DEF(len) F77_CHAR_ARG_USE(s) F77_CHAR_ARG_LEN_USE(s, len)
Use the following macros for Fortran types in C++ code:
F77_INT4
Equivalent to Fortran INTEGER*4
type
F77_DBLE
Equivalent to Fortran DOUBLE PRECISION
type
F77_REAL
Equivalent to Fortran REAL
type
F77_CMPLX
Equivalent to Fortran COMPLEX
type
F77_DBLE_CMPLX
Equivalent to Fortran DOUBLE COMPLEX
type
F77_LOGICAL
Equivalent to Fortran LOGICAL
type
F77_RET_T
Return type of a C++ function that acts like a Fortran subroutine.
Use the following macros to return from C-language functions that are supposed
to act like Fortran subroutines. F77_NORETURN
is intended to be used as
the last statement of such a function that has been tagged with a
"noreturn"
attribute.
F77_RETURN(retval) F77_NORETURN(retval)
The underlying Fortran code should use the XSTOPX
function to replace
the Fortran STOP
function. XSTOPX
uses the Octave exception
handler to treat failing cases in the Fortran code explicitly. Note that
Octave supplies its own replacement BLAS XERBLA
function, which
uses XSTOPX
.
The following example shows the inclusion of a Fortran function in an oct-file, where the C++ wrapper is
#include <octave/oct.h> #include <octave/f77-fcn.h> extern "C" { F77_RET_T F77_FUNC (fortransub, FORTRANSUB) (const F77_INT&, F77_DBLE*, F77_CHAR_ARG_DECL F77_CHAR_ARG_LEN_DECL); } DEFUN_DLD (fortrandemo, args, , "Fortran Demo") { if (args.length () != 1) print_usage (); NDArray a = args(0).array_value (); double *av = a.fortran_vec (); octave_idx_type na = a.numel (); OCTAVE_LOCAL_BUFFER (char, ctmp, 128); F77_FUNC (fortransub, FORTRANSUB) (na, av, ctmp F77_CHAR_ARG_LEN (128)); return ovl (a, std::string (ctmp)); }
and the Fortran function is
subroutine fortransub (n, a, s) implicit none character*(*) s real*8 a(*) integer*4 i, n, ioerr do i = 1, n if (a(i) .eq. 0d0) then call xstopx ('fortransub: divide by zero') else a(i) = 1d0 / a(i) endif enddo write (unit = s, fmt = '(a,i3,a,a)', iostat = ioerr) $ 'There are ', n, $ ' values in the input vector', char(0) if (ioerr .ne. 0) then call xstopx ('fortransub: error writing string') endif return end
This example demonstrates most of the features needed to link to an external Fortran function, including passing arrays and strings, as well as exception handling. Both the Fortran and C++ files need to be compiled in order for the example to work.
mkoctfile fortrandemo.cc fortransub.f [b, s] = fortrandemo (1:3) ⇒ b = 1.00000 0.50000 0.33333 s = There are 3 values in the input vector [b, s] = fortrandemo (0:3) error: fortrandemo: fortransub: divide by zero