Arch Linux User Repository

Looks like I'm the first one to have problems relating to Python. After manually installing ospray and openvr, I'm not a point where I get the following error message: -- Could NOT find Qt3 (missing: QT_QT_LIBRARY QT_INCLUDE_DIR)

ElmerGUI/CMakeLists.txt:93 (FIND_PACKAGE)

Anyone know how to handle this?

Was able to compile by adding this line to the prepare() section of the PKGBUILD

sed -i 's/sprintf(str_pstr, fmt);/sprintf(str_pstr, "%s", fmt);/g' matc/src/str.c

@phollox: same error. It is strange as matc and in particular str.c is present also in other parts of elmer. And they don't give error

Trying to build it. Got this far:

[  3%] Building C object matc/src/CMakeFiles/matc.dir/str.c.o
/home/phollox/.cache/yay/elmerfem/src/elmerfem-release-9.0/matc/src/str.c: In function ‘str_sprintf’:
/home/phollox/.cache/yay/elmerfem/src/elmerfem-release-9.0/matc/src/str.c:85:5: error: format not a string literal and no format arguments [-Werror=format-security]
   85 |     sprintf(str_pstr, fmt);
      |     ^~~~~~~
cc1: some warnings being treated as errors
make[2]: *** [matc/src/CMakeFiles/matc.dir/build.make:328: matc/src/CMakeFiles/matc.dir/str.c.o] Error 1
make[1]: *** [CMakeFiles/Makefile2:14490: matc/src/CMakeFiles/matc.dir/all] Error 2
make: *** [Makefile:166: all] Error 2
==> ERROR: A failure occurred in build().
    Aborting...
error making: elmerfem

Hey, at least is not gfortran complaining....

In case anyone is wondering if the package builds without intervention, as of today the answer is still no.

@hacksd, you're right, the package cannot be built due to a bug in gcc-fortran. It has already been fixed in the development branch of GCC, and I filed a request to maintainers to apply corresponding patch in archlinux. Suggest to wait a bit, or as a temporary solution you may use a different compiler until packages being fixed.

But even after that, I faced multiple issues some of which are reported in the GitHub repo. Finally building from source is what worked for me.

The release used by this package, though being latest, can't be built with gfortran 10 so should this package be flagged as out-of-date?

I tackled that issue with the help of the contributors in this issue. I basically added this sed command in the PKGBUILD before the cmake command:

sed -i -e 's/{CMAKE_Fortran_FLAGS}/{CMAKE_Fortran_FLAGS} -fallow-argument-mismatch/g' ../$pkgname-release-$pkgver/CMakeLists.txt

Hello, thanks for this package. I am facing this error when trying to build this package:

    [  3%] Building Fortran object mathlibs/src/arpack/CMakeFiles/arpack.dir/cnaitr.f.o
/home/sudhanshu/builds/elmerfem/src/elmerfem-release-8.4/mathlibs/src/arpack/cnaitr.f:666:35:

  383 |             call svout (logfil, 1, rnorm, ndigit,
      |                                   2
......
  666 |             call svout (logfil, 2, rtemp, ndigit,
      |                                   1
Error: Rank mismatch between actual argument at (1) and actual argument at (2) (scalar and rank-1)

Any help is welcome.

@CrocoDuck, hi!
I've added a commented check() section to a PKGBUILD, unfortunately, only 96% of tests are passed at the moment, hopefully I will tackle the rest in the next release. None of tests are failing with SIGSEGV, so I'd recommend you (a) to review compiler flags you're passing while building. It could be done in /etc/makepkg.conf, and (b) to look at dependencies of elmer, maybe some of them are out of date on your PC and this produces errors. Also compiling the package with -g option followed by examination of run with valgrind could help to locate the source of the issue.