Problems with external petsc-3.23

[sassy-crick]

Dear all,

I am trying to install fluidity using more modern software packages, which is basically the foss/2024b toolchain from EasyBuild. In particular, and after some patching of the configure files, the build fails in relation to PETSC-3.23.5

mpifort -I../include -J../include -I/home/sassy/apps/software/FFTW/3.3.10-GCC-13.3.0/include -I/home/sassy/apps/software/FlexiBLAS/3.4.4-GCC-13.3.0/include -I/home/sassy/apps/software/FlexiBLAS/3.4.4-GCC-13.3.0/include/flexiblas -I/home/sassy/apps/software/Python/3.12.3-GCCcore-13.3.0/include/python3.12 -DHAVE_NUMPY -I/home/sassy/apps/software/SciPy-bundle/2024.05-gfbf-2024a/lib/python3.12/site-packages/numpy/core/include  -I/home/sassy/easybuild/fluidity/source/new/git/fluidity/include/ -DHAVE_PETSC  -I/home/sassy/apps/software/libspatialindex/2.0.0-GCCcore-13.3.0/include/spatialindex/ -I/home/sassy/apps/software/VTK/9.3.1-foss-2024a/include/vtk-9.3 -DHAVE_VTK -ffast-math -frecord-marker=4 -O2 -ftree-vectorize -march=native -fno-math-errno -I/home/sassy/apps/software/PETSc/3.23.5-foss-2024a/include -I/usr/local/include -I/usr/include -DHAVE_NUMPY -I/home/sassy/apps/software/SciPy-bundle/2024.05-gfbf-2024a/lib/python3.12/site-packages/numpy/core/include -fopenmp -fPIC -fno-realloc-lhs -ffree-line-length-none -ffixed-line-length-none  -O3  -I/home/sassy/easybuild/fluidity/source/new/git/fluidity/include/ -fdefault-real-8 -fdefault-double-8  -I../libwm -I/usr/local/include -I./   -c Petsc_Tools.F90
Petsc_Tools.F90:1337:39:

 1337 |     entries=matrixinfo(MAT_INFO_NZ_USED)
      |                                       1
Error: Symbol ‘mat_info_nz_used’ at (1) has no IMPLICIT type
Petsc_Tools.F90:1328:45:

 1328 |     double precision, dimension(MAT_INFO_SIZE):: matrixinfo
      |                                             1
Error: Symbol ‘mat_info_size’ at (1) has no IMPLICIT type
Petsc_Tools.F90:1336:50:

 1336 |     call MatGetInfo(matrix, MAT_LOCAL, matrixinfo)
      |                                                  1
Error: There is no specific subroutine for the generic ‘matgetinfo’ at (1)
Petsc_Tools.F90:1385:73:

 1385 |         call MatGetRow(matrix, offset+i, ncols, row_cols, row_vals, ierr)
      |                                                                         1
Error: There is no specific subroutine for the generic ‘matgetrow’ at (1)
Petsc_Tools.F90:1402:77:

 1402 |         call MatRestoreRow(matrix, offset+i, ncols, row_cols, row_vals, ierr)
      |                                                                             1
Error: There is no specific subroutine for the generic ‘matrestorerow’ at (1)
Petsc_Tools.F90:1413:83:

 1413 |         call MatGetRow(matrix, offset+i, ncols, sparsity%colm(j:), A%val(j:), ierr)
      |                                                                                   1
Error: There is no specific subroutine for the generic ‘matgetrow’ at (1)
Petsc_Tools.F90:1417:87:

 1417 |         call MatRestoreRow(matrix, offset+i, ncols, sparsity%colm(j:), A%val(j:), ierr)
      |                                                                                       1
Error: There is no specific subroutine for the generic ‘matrestorerow’ at (1)
Petsc_Tools.F90:1116:45:

 1116 |       0, nnz, 0, PETSC_NULL_INTEGER, M, ierr)
      |                                             1
Error: There is no specific subroutine for the generic ‘matcreateaij’ at (1)
make: *** [Makefile:124: Petsc_Tools.o] Error 1

when I am executing make in the femtools directory.
This also happens when I am using the GitHub branch, so not only when I am using the latest release version what I am normally doing.

Are there any plans to upgrade the software so it can use more modern software packages like the mentioned PETSc and also the issue around how it finds VTK ?

[Patol75]

Hi @sassy-crick,

Thank you for the report. Yes, there are finally plans to support building Fluidity with recent dependency versions officially. Much work toward it is already done, but we still need to finalise some aspects. We will keep you updated in the coming weeks.

[Patol75]

#390 was just merged, which should allow compatibility with more recent PETSc and VTK library versions.

[Patol75]

I am closing this issue, as #390, #391, and #392 should address it.