This is the mail archive of the 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]

[newbie] "linker input file unused because linking not done"?

I'm guessing that either I'm making an embarrassingly simple mistake
(my background is nearly all with VMs and interpreters), or that the
problem below is somehow related to my converting a build process
accustomed to using icc/ifc to gcc/gfortran. In any case, I'm stumped:

I have a Makefile with

> FC     = gfortran
> FFLAGS = -ffixed-line-length-132 -w -c -m64 -march=native -funroll-loops --param max-unroll-times=4 -ftree-vectorize -ftree-vectorizer-verbose=0 -O2 -I$(NETCDF)/include -I$(IOAPI_ROOT)/src -I.
> LIBS   = -L$(IOAPI_ROOT)/lib -lioapi \
>          -L$(NETCDF)/lib -lnetcdff -lnetcdf

followed by lists of MODULES and OBJS, then

> all:
>       @$(MAKE) $(MODULES)
>       @$(MAKE) $(MODEL)

> $(MODEL): $(OBJS)
>       $(FC) -o $(MODEL) $(FFLAGS) $(OBJS) $(MODULES) $(LIBS)


`make` compiles 96 fortran sources to object files without complaint,
then runs

> gfortran -o runme <$FFLAGS expanded here/> runme.o <95 more *.o here/> <$LIBS expanded here/>

which produces 96 lines like

> gfortran: warning: <object filename here/>: linker input file unused because linking not done

What must I utter to cause the *.o (listed in the linkage command in
the Makefile) in the current directory to be linked?

FWIW I also tried FFLAGS ending with '-L.' instead of 'I.': no change.

I suspect I need to Read Some Fine Manual: feel free to point me to one.

TIA, Tom Roche <>

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