Compline on macOS 10.15.6

FreeFEM installation
1

I initially pulled from git a couple of months back and had no trouble building. After updating to macOS 10.15.6 I’m seeing the error below. Unfortunately, I also pulled fresh sources for freeFEM. Thus the sources may have changed, or the updated build tools (supplemented with Home-brew) may have changed.

cd MUMPS_5.0.2 && /Applications/Xcode.app/Contents/Developer/usr/bin/make d
/Applications/Xcode.app/Contents/Developer/usr/bin/make ARITH=d mumps_lib
(cd libseq; /Applications/Xcode.app/Contents/Developer/usr/bin/make)
gfortran -O -g -DNDEBUG -O3 -O3 -O3 -fPIC -c mpi.f -o mpi.o
mpi.f:566:72:

566 | CALL MUMPS_COPY_LOGICAL( SENDBUF, RECVBUF, COUNT )
| 1
Error: Type mismatch in argument ‘s’ at (1); passed INTEGER(4) to LOGICAL(4)
mpi.f:568:72:

568 | CALL MUMPS_COPY_REAL( SENDBUF, RECVBUF, COUNT )
| 1
Error: Type mismatch in argument ‘s’ at (1); passed INTEGER(4) to REAL(4)
mpi.f:571:72:

571 | CALL MUMPS_COPY_DOUBLE_PRECISION( SENDBUF, RECVBUF, COUNT )
| 1
Error: Type mismatch in argument ‘s’ at (1); passed INTEGER(4) to REAL(8)
mpi.f:573:72:

573 | CALL MUMPS_COPY_COMPLEX( SENDBUF, RECVBUF, COUNT )
| 1
Error: Type mismatch in argument ‘s’ at (1); passed INTEGER(4) to COMPLEX(4)
mpi.f:575:72:

575 | CALL MUMPS_COPY_DOUBLE_COMPLEX( SENDBUF, RECVBUF, COUNT )
| 1
Error: Type mismatch in argument ‘s’ at (1); passed INTEGER(4) to COMPLEX(8)
mpi.f:577:72:

577 | CALL MUMPS_COPY_2DOUBLE_PRECISION( SENDBUF, RECVBUF, COUNT )
| 1
Error: Type mismatch in argument ‘s’ at (1); passed INTEGER(4) to REAL(8)
make[8]: *** [mpi.o] Error 1
make[7]: *** [libseqneeded] Error 2
make[6]: *** [d] Error 2
make[5]: *** [MUMPS_5.0.2/FAIT] Error 2
make[4]: *** [compile-dir] Error 2
make[3]: *** [tag-compile-pkg] Error 1
make[2]: *** [all-recursive] Error 1
make[1]: *** [all-recursive] Error 1
make: *** [all] Error 2

2

Apologies. I see this is a problem on MUMPS. Possibly due to an update to gfortran. Adding the gfortran option -fallow-argument-mismatch allow mpi.f to compile

3

You can also let PETSc compile MUMPS with make petsc-slepc, and then ./reconfigure --disable-fortran.

4

Hi
Same problem for me.

The ./reconfigure --disable-fortran method did not work for me.

As for the other method, can you explain me how you changed the gfortran options to compile correctly ?

Thanks

5

Things are getting even worse…

To make sure to redo every steps of the compilations I did a
make clean and then make petsc-slepc in the ff-petsc directory and now it fails to recompute slepc !

And the error message is of very little help (see bellow)

HELP PLEASE !
David

=============================================================================== Compiling and installing Scalapack; this may take several minutes =============================================================================== *******************************************************************************
UNABLE to CONFIGURE with GIVEN OPTIONS (see configure.log for details):

Error running make on SCALAPACK

make: *** [petsc-3.13.0/tag-conf-real] Error 1
(base) fabred@autonome232:ff-petsc$ more config.log

----------------------

Running config.status.

----------------------

This file was extended by FreeFEM config.status 4.6, which was
generated by GNU Autoconf 2.69. Invocation command line was

CONFIG_FILES =
CONFIG_HEADERS =
CONFIG_LINKS =
CONFIG_COMMANDS =
$ …/…/config.status --file=Makefile.inc:Makefile-PETSc.inc

on autonome232.local

config.status:1224: creating Makefile.inc

6

Nothing to do with SLEPc… It says ScaLAPACK, not SLEPc. You need to switch to the develop branch.

7

OK. So switched to develop branch.
I had to do ./reconfigure --disable-fortran to succeed compilation of mpi.f

Now compilation succeeds but new problem:

(base) fabred@autonome232:FreeFem-sources$ src/mpi/FreeFem++-mpi

src/mpi/FreeFem++-mpi dyld: Library not loaded: @rpath/libhdf5.103.dylib Referenced from: /Users/fabred/FreeFem-sources/src/mpi/FreeFem++-mpi Reason: image not found Abort trap: 6

Thanks for help !

EDIT : problem finally fixed

using develop branch with configure options

--disable-fortran --disable-iohdf5 --with-hdf5=no --without-hdf5

8

A quick search on the forum returns this solution.

9

Just for completeness … I am able to compile by manually adding “-fallow-argument-mismatch” when needed.

I’ll try the configure options and report back.

10

I’m able to build with David’s suggested configure change. I get two FAIL during “make check” and will follow up with another thread.