This is the mail archive of the mailing list for the GCC 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]

getarg_ and iargc_


So this Mandriva guy tells me gfortran can't compile Scilab, and
he is right!  Scilab is a pretty important piece of Fortran that
many people use, so it is a shame that gfortran can't build it
right now.  But the reason why it can't is not really a Fortran
problem (at least I haven't found any yet ;-) but a compatibility
problem with the library wrt. what g77 had (libU77?)...

Scilab uses some kind of C <-> Fortran compatiblity layer where
they have a getarg.c that calls C2F(getarg), with a line in one
of the included header files that says "#define C2F(name) name##_".
The same thing happens with iargc, so the gerarg.o object that
the linker sees ends up calling getarg_ and iargc_.

This worked Just Fine with g77, and apparently it works with g95
as well.  It does not work with gfortran, because we have instead
gfortran_getarg_i4 and so on.

I'm looking for suggestions for how to fix this in libgfortran.
I would fix this in Scilab, except that if gfortran is to be a
drop-in replacement for g77, then we shouldn't have this kind of
rather silly compatibilty issues.



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