jlokimlin / fishpack Goto Github PK

Hi everybody!
I can not build fishpack using both Ubuntu 18.04.2 LTS (gfortran 7.3.0) and Manjaro 18.0.4 (gfortran 8.2.1) with the same error.

What should I do to enjoy the fishpack?

Everything is OK until following stage:
gfortran -O3 -std=f2008ts -fimplicit-none -Wall -c staggered_axisymmetric_spherical_solver.f90 -I../objs -I../lib \
 -o ../objs/staggered_axisymmetric_spherical_solver.o -J../lib
*** buffer overflow detected ***: /usr/lib/gcc/x86_64-linux-gnu/7/f951 terminated
f951: internal compiler error: Aborted
Please submit a full bug report,
with preprocessed source if appropriate.
See <file:///usr/share/doc/gcc-7/README.Bugs> for instructions.
Makefile:118: recipe for target 'staggered_axisymmetric_spherical_solver.o' failed
make[2]: *** [staggered_axisymmetric_spherical_solver.o] Error 1
make[2]: Leaving directory '/home/jack/Downloads/fishpack/src'
/bin/sh: 1: ./build_lib.exe: not found
Makefile:250: recipe for target 'all' failed
make[1]: *** [all] Error 127
make[1]: Leaving directory '/home/jack/Downloads/fishpack/src'
Makefile:9: recipe for target 'lib' failed
make: *** [lib] Error 2

If I ignore the module staggered_axisymmetric_spherical_solver.f90, then the same error occurs at next module compiling, and so on.

Hi there, thanks for the great tool. I want to call the library within windows, so I need to compile the library into a dll.

I created a build pipeline with CMake, and am now trying to compile with Codeblocks using gfortran. I get the following error:

I:\Software\fishpack-master\src\centered_helmholtz_solvers.f90:37:8:
module subroutine hwscrt(a, b, m, mbdcnd, bda, bdb, c, d, n, nbdcnd, bdc, &
1
Error: Unclassifiable statement at (1)

Do you perhaps have any suggestion for how to fix this problem? Or have you ever tried compiling for a windows machine? Thanks for any help.

I am wondering why the assertion assert_equal is defined as TRUE when abs(known_error - discretization_error) > 0. This allows all examples to PASS, when in fact, the difference is non-zero.

It seems that abs(known_error - discretization_error) < tol, for some small value tol (e.g. tol = 1e-12) would make more sense.