Summary: | sci-libs/hdf5-1.8.14-r1 - mpif90 .../work/hdf5-1.8.14/fortran/testpar/ptest.f90: mpif-config.h:6: Illegal preprocessor directive - Error: Invalid character in name at (1) | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Juergen Rose <rose> |
Component: | [OLD] Library | Assignee: | Kacper Kowalik (Xarthisius) (RETIRED) <xarthisius> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | frp.bissey, sci |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
See Also: |
https://github.com/gentoo-science/sci/issues/372 https://bugs.gentoo.org/show_bug.cgi?id=519700 |
||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
Juergen Rose
2015-02-18 14:17:31 UTC
BTW., /usr/include/x86_64-pc-linux-gnu/mpif-config.h comes from openmpi-1.8.4-r2: root@impala:/root(55)# genlop -t openmpi | tail Wed Feb 4 12:27:58 2015 >>> sys-cluster/openmpi-1.8.4 merge time: 8 minutes and 57 seconds. Fri Feb 6 08:53:00 2015 >>> sys-cluster/openmpi-1.8.4-r1 merge time: 15 minutes and 21 seconds. Tue Feb 17 08:57:04 2015 >>> sys-cluster/openmpi-1.8.4-r2 merge time: 11 minutes and 48 seconds. openmpi-1.8.4-r2 came form the science overlay; After masking >=openmpi-1.8.4-r2 and reinstalling openmpi-1.8.4-r1 the next 'emerge -ptv1 hdf5' wants to remove openmpi completely and install sys-cluster/mpich: root@impala:/root(63)# MAKEOPTS=-j1 emerge -ptv1 hdf5 These are the packages that would be merged, in reverse order: Calculating dependencies... done! [ebuild R ] sci-libs/hdf5-1.8.14-r1:0/1.8.14::gentoo USE="examples fortran mpi szip zlib -cxx -debug -fortran2003 -static-libs -threads" 0 KiB [ebuild N ] virtual/mpi-2.0-r4::science USE="cxx fortran romio threads" ABI_X86="(64) -32 (-x32)" 0 KiB [nomerge ] sci-mathematics/freemat-4.2::gentoo USE="volpack vtk" [nomerge ] sci-libs/vtk-6.1.0-r1::gentoo USE="R X boost cg doc ffmpeg gdal java mpi mysql odbc postgres python qt4 rendering smp tbb tcl theora tk -all-modules (-aqua) -examples -imaging -json -kaapi -offscreen {-test} -views -web -xdmf2" PYTHON_TARGETS="python2_7" VIDEO_CARDS="-nvidia" [nomerge ] virtual/mpi-2.0-r4::science USE="cxx fortran romio threads" ABI_X86="(64) -32 (-x32)" [nomerge ] sys-cluster/mpich-3.1.3-r1::science USE="cxx doc fortran romio threads -mpi-threads" ABI_X86="(64) -32 (-x32)" [blocks b ] sys-cluster/openmpi ("sys-cluster/openmpi" is blocking sys-cluster/mpich-3.1.3-r1) [uninstall ] sys-cluster/openmpi-1.8.4-r1::gentoo USE="cxx fortran ipv6 romio threads -cma -cuda -heterogeneous -mpi-threads -vt" OPENMPI_FABRICS="-knem -ofed -open-mx -psm" OPENMPI_OFED_FEATURES="-connectx-xrc -control-hdr-padding -dynamic-sl -failover -rdmacm -udcm" OPENMPI_RM="-pbs -slurm" [ebuild N ] sys-cluster/mpich-3.1.3-r1::science USE="cxx doc fortran romio threads -mpi-threads" ABI_X86="(64) -32 (-x32)" 11,219 KiB I would appreciate any hint, how I can handle this issue. I do not want to install mpich. If I also mask sys-cluster/mpich and sys-cluster/mpich2, 'emerge -ptv1 hdf5' wants now to install sys-cluster/mvapich2. But at least it will not to remove openmpi: root@impala:/root(67)# MAKEOPTS=-j1 emerge -ptv1 hdf5 These are the packages that would be merged, in reverse order: Calculating dependencies... done! [ebuild R ] sci-libs/hdf5-1.8.14-r1:0/1.8.14::gentoo USE="examples fortran mpi szip zlib -cxx -debug -fortran2003 -static-libs -threads" 0 KiB [ebuild N ] virtual/mpi-2.0-r4::science USE="cxx fortran romio threads" ABI_X86="(64) -32 (-x32)" 0 KiB [ebuild N ] sys-cluster/mvapich2-1.6-r1::science USE="fortran romio threads -debug -large-cluster -medium-cluster" 22,696 KiB [ebuild N ] sys-infiniband/libibmad-1.3.11:3.12::gentoo 21,539 KiB [ebuild N ] sys-infiniband/librdmacm-1.0.18:3.12::gentoo 0 KiB [ebuild N ] sys-infiniband/libibumad-1.3.9:3.12::gentoo 0 KiB [ebuild N ] sys-infiniband/libibverbs-1.1.7:3.12::gentoo 0 KiB And 'MAKEOPTS=-j1 emerge -tv1 hdf5' then fails emerging sys-cluster/mvapich2. Why does sci-libs/hdf5 no more work with openmpi? I don't need this infiniband stuff. Removing the the mpi USE flag from hdf5 does not help either. If I try this I get slot conflict with several other packages: root@impala:/root(71)# MAKEOPTS=-j1 emerge -ptv1 hdf5 These are the packages that would be merged, in reverse order: Calculating dependencies... done! [ebuild R ] sci-libs/hdf5-1.8.14-r1:0/1.8.14::gentoo USE="examples fortran szip zlib -cxx -debug -fortran2003 -mpi* -static-libs -threads" 0 KiB Total: 1 package (1 reinstall), Size of downloads: 0 KiB !!! Multiple package instances within a single package slot have been pulled !!! into the dependency graph, resulting in a slot conflict: sci-libs/hdf5:0 (sci-libs/hdf5-1.8.14-r1:0/1.8.14::gentoo, ebuild scheduled for merge) pulled in by sci-libs/hdf5 (Argument) (sci-libs/hdf5-1.8.14-r1:0/1.8.14::gentoo, installed) pulled in by sci-libs/hdf5:0/1.8.14=[mpi,szip,zlib] required by (sci-libs/netcdf-4.3.2-r1:0/7::gentoo, installed) ^^^ sci-libs/hdf5:0/1.8.14=[mpi] required by (dev-python/h5py-2.4.0:0/0::gentoo, installed) ^^^ sci-libs/hdf5:=[mpi=] required by (dev-python/h5py-2.4.0:0/0::gentoo, installed) ^^^^ >=sci-libs/hdf5-1.8.0:=[mpi?] required by (media-libs/vigra-1.10.0:0/0::gentoo, installed) ^^^^ sci-libs/hdf5:0=[mpi=,szip=,zlib] required by (sci-libs/netcdf-4.3.2-r1:0/7::gentoo, installed) ^^^^ sci-libs/hdf5[mpi=] required by (sci-visualization/paraview-4.2.0-r1:0/0::gentoo, installed) ^^^^ It might be possible to solve this slot collision by applying all of the following changes: - sci-libs/hdf5-1.8.14-r1 (Change USE: +mpi) Nothing is changed with [science-overkay]sys-cluster/openmpi-1.8.4-r3. I have been discussing this problem as a github issue https://github.com/gentoo-science/sci/issues/372 look it up for an explanation of what's going on. I added a workaround, please re-emerge openmpi. (In reply to Christoph Junghans from comment #8) > I added a workaround, please re-emerge openmpi. Thanks Christoph for the workaround, it works now for me. |