Compilation feedback

[Kevin]

Hi all,

I am installing He-3 and ran into a few compilation issues - small
issues, but took me some time to figure out.


First though - I'm not so happy with the feedback mechanisms.

** A forum AND a mailing list? Confusing -> several places to look
for solutions or report problems. Nowhere on the forum does it say
there's a mailing list, too.

** I have posted in the forum before, but today I could not: I got an
error message saying I needed an account. So I make an account (the
last thing I need is another account!), and then I'm a low-avatar user
or some bullshit and I still can't submit my post. (It's pasted
below, so you can judge for yourself if there's anything offensive in
it). Then I try to look for "contact" options - a mailing list ...


I'M DOING *YOU* A FAVOR BY SUBMITTING A BUG REPORT - PLEASE MAKE IT
EASY ON ME - E.G. A NO B.S. WEBFORM THAT DOESN'T REQUIRE ME TO SIGN UP
OR TRACK YOU DOWN.

** The "problem flags" I am reporting below were already reported by
someone on the forum in October 2011, 5 months ago ... It wasn't
really necessary for me to have to spend 2 hours fixing things
tonight.


Thank you for your consideration!



( Trying to post in response to http://exciting-code.org/forum/t-306579/successful-installation-on-linux-ifort
: )


1. I'm confirming the observation made by Ralf for exciting-he-3.
Furthermore, the "rescaleYS" error occurs for all or almost all tests,
not just test05.

2. However, on my system (using ifort 11.1.075), I also need to
remove -check all -warn all to be able to run "make" in the test
folder. Otherwise, I failt to compile with errors such as :

ifort -g -O0 -warn all -check all -traceback -ftrapuv -cpp -DXS -DISO
-DTETRA -DLIBXC -I./finclude -c ../../src/fftlib/cfftnd.f90
../../src/fftlib/cfftnd.f90(470): error #6633: The type of the actual
argument differs from the type of the dummy argument. [C]
call cmf2kb ( lot, ido, l1, na, c, jump, inc, ch, 1, lot,
wa(iw) )
--------------------------------------^



3. I note that, after editing build/make.inc and setting
F90/77_DEBUGOPTS = -g -O0 -traceback, and doing "make cleantests" in
test/ , I still had to go to build/debug, do "make clean", and then go
back to test/ and do "make" in order for everything to be rebuilt
without the offensive -ftrapuv.


4. There is a misnamed file :
mv test02/runlapackmpi/srartmpirun test02/runlapackmpi/startmpirun


I would be happy if the authors could remove these problematic options
from make.inc....


Cheers

Kevin



[root@ip-10-17-19-252 exciting-he3-install]# more build/make.inc
# Using ifort 11.1.075+mkl, and openmpi-1.4.3
F90=ifort
F90_OPTS = -O3 -ip -unroll -scalar_rep -w90
CPP_ON_OPT = -cpp -DXS -DISO -DTETRA -DLIBXC
F77=$(F90)
F77_OPTS = -O3
LIB_ARP =libarpack.a
#export USE_SYS_LAPACK=true
LIB_LPK = -L/opt/intel/mkl/10.0.1.014/lib/em64t/ -lmkl_lapack -
lmkl_core -lmkl -liomp5 -lpthread
#MKLROOT = /local/intel64/Compiler/11.1/075/mkl/lib/em64t/
LIB_LPK = $(MKLROOT)/libmkl_blas95_lp64.a $(MKLROOT)/
libmkl_lapack95_lp64.a $(MKLROOT)/libmkl_scalapack_lp64.a -Wl,--start-
group $(MKLROOT)/libmkl_intel_lp64.a $(MKLROOT)/libmkl_sequential.a $
(MKLROOT)/libmkl_core.a $(MKLROOT)
/libmkl_blacs_openmpi_lp64.a -Wl,--end-group -lpthread -lm
LIB_FFT = fftlib.a
LIB_BZINT=libbzint.a

LIBS= $(LIB_ARP) $(LIB_LPK) $(LIB_FFT) $(LIB_BZINT)

F90_DEBUGOPTS=-g -O0 -warn all -check all -traceback -ftrapuv
F77_DEBUGOPTS=-g -O0 -warn all -check all -traceback -ftrapuv
F90_DEBUGOPTS=-g -O0 -traceback
F77_DEBUGOPTS=-g -O0 -traceback

#Ignore if you don't have MPI or smplibs

MPIF90=mpif90
MPIF90_OPTS=$(F90_OPTS) $(CPP_ON_OPT) -DMPI -DMPIRHO -DMPISEC

F77MT=$(F77)
F90MT=$(F90)
SMP_LIBS= $(LIB_ARP) $(LIB_LPK) $(LIB_FFT) $(LIB_BZINT)
SMPF90_OPTS=$(F90_OPTS)
SMPF77_OPTS=$(SMPF90_OPTS)
MPISMPF90_OPTS=$(SMPF90_OPTS) -DMPI -DMPIRHO -DMPISECBUILDMPI=false
MPIF90MT=$(MPIF90)
BUILDMPI=true
BUILDSMP=false