Hi, Ludovic Courtès <[email protected]> writes:
> MIA 2.4.6 fails to build as of 9098745b181b3022587a35afd255f7ff1d41ac86 > presumably due to CMake shenanigans as reported here: > > https://issues.guix.gnu.org/46724#9 > > Build failure looks like this: > > [ 13%] Building CXX object > mia/core/CMakeFiles/test-core.dir/test_core_combined.cc.o > cd /tmp/guix-build-mia-2.4.6.drv-0/build/mia/core && > /gnu/store/rn75fm7adgx3pw5j8pg3bczfqq1y17lk-gcc-7.5.0/bin/c++ > -DBOOST_ALL_DYN_LINK -DBOOST_ALL_NO_LIB -DBOOST_FILESYSTEM_DYN_LINK > -DBOOST_FILESYSTEM_VERSION=3 -DBOOST_SYSTEM_DYN_LINK > -DBOOST_UNIT_TEST_FRAMEWORK_DYN_LINK -DHAVE_CONFIG_H > -I/gnu/store/n50ikpnyzdda0zklwnmqasklcrif2csg-eigen-3.3.8/include/eigen3 > -I/gnu/store/c8w9z48vvx2a3q3k44ch9yn00wk1qwhb-libxml2-2.9.10/include/libxml2 > -I/tmp/guix-build-mia-2.4.6.drv-0/build > -I/tmp/guix-build-mia-2.4.6.drv-0/mia-2.4.6 -fpermissive > -fvisibility=hidden -O2 -g -DNDEBUG -o > CMakeFiles/test-core.dir/test_core_combined.cc.o -c > /tmp/guix-build-mia-2.4.6.drv-0/mia-2.4.6/mia/core/test_core_combined.cc > /tmp/guix-build-mia-2.4.6.drv-0/mia-2.4.6/addons/vtk/vtkmesh.cc: In function > ‘void vtkmia::read_normals(mia::CTriangleMesh&, vtkPointData&)’: > /tmp/guix-build-mia-2.4.6.drv-0/mia-2.4.6/addons/vtk/vtkmesh.cc:179:12: > error: ‘class vtkFloatArray’ has no member named ‘GetTupleValue’; did you > mean ‘GetValue’? > normals->GetTupleValue(i, &is->x); > ^~~~~~~~~~~~~ > GetValue > /tmp/guix-build-mia-2.4.6.drv-0/mia-2.4.6/addons/vtk/vtkmesh.cc: In function > ‘void vtkmia::read_colors(mia::CTriangleMesh&, vtkPointData&)’: > /tmp/guix-build-mia-2.4.6.drv-0/mia-2.4.6/addons/vtk/vtkmesh.cc:211:11: > error: ‘class vtkFloatArray’ has no member named ‘GetTupleValue’; did you > mean ‘GetValue’? > colors->GetTupleValue(i, &is->x); > ^~~~~~~~~~~~~ > GetValue > make[2]: *** [addons/vtk/CMakeFiles/mesh-io-vtkmesh-common.dir/build.make:66: > addons/vtk/CMakeFiles/mesh-io-vtkmesh-common.dir/vtkmesh.cc.o] Error 1 > make[2]: Leaving directory '/tmp/guix-build-mia-2.4.6.drv-0/build' > > The problem here is what we’re missing -I flags for VTK, which might be > a CMake incompatibility issue. Our 2.4.7 was failing again, apparently due to a newer hdf5 and GCC, so I've applied patches from Debian among other tweaks in commit d9831cd4399b0f81a8bd4eddab1a0c075bc7751f. Sadly, its doc is currently non-deterministic, see bug#78012. Anyway, closing this 4 yo one. -- Thanks, Maxim
