Fix processing of CMake option `with-mpi` #3112

gtrensch · 2024-02-23T08:48:55Z

Meant to address #2993, but does not.

muffgaga · 2024-02-23T09:18:59Z

Looks good to me, should I perform a test build in my setup or do you already have something in place test-build-wise?

heplesser · 2024-02-23T09:28:39Z

@muffgaga I think it would be good if you could test it on your actual system.

@gtrensch Did you have a look through the ProcessOptions.cmake file to see if there are other options that would benefit from the same fix or are they all correctly written from before?

terhorstd · 2024-02-26T15:51:40Z

I think this also fixes #2179, right?

muffgaga · 2024-02-26T16:53:53Z

@muffgaga I think it would be good if you could test it on your actual system.

I did recreate my initial error (sorry, somehow deleted my state too early to ;-) ) using the current EBRAINS state for the nest package (and spack) → building nest@3.6~mpi+sonata; I then added the diff as a patch, but unfortunately it still looks wrong:

-- Checking for module 'mpi-c'
--   Package 'mpi-c', required by 'virtual:world', not found
-- Checking for module 'mpi-cxx'
--   Package 'mpi-cxx', required by 'virtual:world', not found
CMake Error at /opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindPackageHandleStandardArgs.cmake:230 (message):
  Could NOT find MPI (missing: MPI_C_FOUND MPI_CXX_FOUND)
Call Stack (most recent call first):
  /opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindPackageHandleStandardArgs.cmake:600 (_FPHSA_FAILURE_MESSAGE)
  /opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindMPI.cmake:1837 (find_package_handle_standard_args)
  /fasthome/mueller/code/ebrains-spack-builds/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/hdf5-1.14.3-wbpbob3l2cdywg4z2ig2xkctyff4qlyc/cmake/hdf5-config.cmake:105 (find_package)
  /opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindHDF5.cmake:486 (find_package)
  cmake/ProcessOptions.cmake:608 (find_package)
  CMakeLists.txt:154 (nest_process_with_hdf5)


-- Configuring incomplete, errors occurred!
==> Error: ProcessError: Command exited with status 1:
    '/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/bin/cmake' '-G' 'Unix Makefiles' '-DCMAKE_INSTALL_PREFIX:STRING=/fasthome/mueller/code/ebrains-spack-builds/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/nest-3.6-a6hjqgoldoqp45mirsovtafvq4pvu6en' '-DCMAKE_BUILD_TYPE:STRING=Release' '-DCMAKE_INTERPROCEDURAL_OPTIMIZATION:BOOL=OFF' '-DCMAKE_VERBOSE_MAKEFILE:BOOL=ON' '-DCMAKE_INSTALL_RPATH_USE_LINK_PATH:BOOL=ON' '-DCMAKE_INSTALL_RPATH:STRING=/fasthome/mueller/code/ebrains-spack-builds/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/nest-3.6-a6hjqgoldoqp45mirsovtafvq4pvu6en/lib;/fasthome/mueller/code/ebrains-spack-builds/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/nest-3.6-a6hjqgoldoqp45mirsovtafvq4pvu6en/lib64;/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/gsl-2.7.1-vsbdfi6jgf3ymt6sxz3emq4s5smjkhlj/lib' '-DCMAKE_PREFIX_PATH:STRING=/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/py-pandas-1.5.3-e5y7lnc4yj4kygwnuugif2qhgfd7i7pm;/fasthome/mueller/code/ebrains-spack-builds/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/py-h5py-3.8.0-j36pwzcof23aktbs7y33pjhnnspzw3on;/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/py-numpy-1.24.3-nszdl2rkpjdw5fkpe2ug34q5h7rvra75;/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/py-cython-0.29.33-n3yd5sru4tjx5smsoexkv7dkbjeycoab;/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/py-setuptools-63.4.3-olvcbntq6mcffo7ulehyp6oyhyunfhjn;/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/python-3.8.2-3qcqdgntmvo42377azozox4rkxi36eah;/fasthome/mueller/code/ebrains-spack-builds/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/hdf5-1.14.3-wbpbob3l2cdywg4z2ig2xkctyff4qlyc;/fasthome/mueller/code/ebrains-spack-builds/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/pkgconf-1.8.0-63ve3u3eb4kgihgoxzypqu6a575d3jst;/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/libtool-2.4.7-ce5pzk62q2owgvpqr6is4mx6iy7krgpf;/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/gsl-2.7.1-vsbdfi6jgf3ymt6sxz3emq4s5smjkhlj;/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/gmake-4.4.1-mkul7cltni62fuhqfrd6utzcyuhxmjjh;/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk;/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/readline-8.2-etwpmxu3ogwyt63zrjnxygcwxmna37wa;/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/ncurses-6.4-gbmndp32rvppfgwuoey25cfgmlwdg2o3;/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/boost-1.79.0-cq4pl77kehytn56ey5zjajcsjzpzlekl' '-DCMAKE_INSTALL_LIBDIR=lib' '-DSPACK_CXX_COMPILER=/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-10.2.1/gcc-11.2.0-dxrcp4vfmgkwad3moqpgkdqyavxz6cxa/bin/g++' '-Dwith-boost=ON' '-Dwith-mpi=OFF' '-Dwith-openmp=ON' '-Dwith-optimize=ON' '-Dwith-gsl=/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/gsl-2.7.1-vsbdfi6jgf3ymt6sxz3emq4s5smjkhlj' '-Dwith-hdf5=ON' '-Dstatic-libraries=OFF' '/tmp/mueller/spack-stage/spack-stage-nest-3.6-a6hjqgoldoqp45mirsovtafvq4pvu6en/spack-src'

1 error found in build log:
     64    -- Found OpenMP: TRUE (found version "4.5")
     65    -- Found Boost: /opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/boost-1.79.0-cq4pl77kehytn56ey5zjajcsjzpzlekl/lib/cmake/Boost-1.79.0/BoostConfig.cmake (found suitable version "1.79.0
           ", minimum required is "1.69.0")
     66    -- Checking for module 'mpi-c'
     67    --   Package 'mpi-c', required by 'virtual:world', not found
     68    -- Checking for module 'mpi-cxx'
     69    --   Package 'mpi-cxx', required by 'virtual:world', not found
  >> 70    CMake Error at /opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindPackageHandleStandardArgs.cmake:230 (message
           ):
     71      Could NOT find MPI (missing: MPI_C_FOUND MPI_CXX_FOUND)
     72    Call Stack (most recent call first):
     73      /opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindPackageHandleStandardArgs.cmake:600 (_FPHSA_FAILURE_MESSA
           GE)
     74      /opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindMPI.cmake:1837 (find_package_handle_standard_args)
     75      /fasthome/mueller/code/ebrains-spack-builds/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/hdf5-1.14.3-wbpbob3l2cdywg4z2ig2xkctyff4qlyc/cmake/hdf5-config.cmake:105 (find_package)
     76      /opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindHDF5.cmake:486 (find_package)

See build log for details:
  /tmp/mueller/spack-stage/spack-stage-nest-3.6-a6hjqgoldoqp45mirsovtafvq4pvu6en/spack-build-out.txt

mueller@helvetica:~/..ebrains-spack-builds (spack_v0.21.1 *) [1] $ grep with-mpi /tmp/mueller/spack-stage/spack-stage-nest-3.6-a6hjqgoldoqp45mirsovtafvq4pvu6en/spack-src/cmake/ProcessOptions.cmake | head -n 1
  if ( NOT "${with-mpi}" STREQUAL "OFF" )

If I build nest@3.6~mpi+sonata but use ^hdf5~mpi (i.e. disable MPI in nest's hdf5 dependency (which is caused by the +sonata variant)) → no detection of mpi happens, and build is fine.

⇒ I guess some cmake file provided by the installed hdf5 triggers the mpi check (and inheriting mpi ~~is probably~~ might be the right thing to do) but it fails.

heplesser · 2024-02-27T13:13:08Z

Just a comment: On my system (macOS 14.3.1) with cmake 3.28.3, -Dwith-mpi=OFF works as advertised. With this setting, the MPI-tests are not performed at all:

-- Found GSL: /Users/plesser/mambaforge/envs/nest/include (found version "2.7") 
-- Found OpenMP_C: -Xclang -fopenmp (found version "5.0") 
-- Found OpenMP_CXX: -Xclang -fopenmp (found version "5.0") 
-- Found OpenMP: TRUE (found version "5.0")  
-- Found Boost: /Users/plesser/mambaforge/envs/nest/lib/cmake/Boost-1.84.0/BoostConfig.cmake (found suitable version "1.84.0", minimum required is "1.69.0")

while they are run when using -Dwith-mpi=ON:

-- Found GSL: /Users/plesser/mambaforge/envs/nest/include (found version "2.7") 
-- Found OpenMP_C: -Xclang -fopenmp (found version "5.0") 
-- Found OpenMP_CXX: -Xclang -fopenmp (found version "5.0") 
-- Found OpenMP: TRUE (found version "5.0")  
-- Found MPI_C: /Users/plesser/mambaforge/envs/nest/lib/libmpi.dylib (found version "3.1") 
-- Found MPI_CXX: /Users/plesser/mambaforge/envs/nest/lib/libmpi.dylib (found version "3.1") 
-- Found MPI: TRUE (found version "3.1")  
-- Found PY_mpi4py: /Users/plesser/mambaforge/envs/nest/lib/python3.12/site-packages/mpi4py  
-- Found Boost: /Users/plesser/mambaforge/envs/nest/lib/cmake/Boost-1.84.0/BoostConfig.cmake (found suitable version "1.84.0", minimum required is "1.69.0")

I am very certain that this has worked like this for many years, so that it is not a matter of a very recent Cmake update.

The same holds true for our NEST CI builds, e.g. the one, only including BOOST, see

where comparison of the last line with the -Dwith-mpi=ON case shows that in the latter case MPI libraries are included when linking NEST:

https://github.com/nest/nest-simulator/actions/runs/8001705514/job/21853452376#step:18:3696

So it seems to me that the ON/OFF flag works in a number of cases at least with the existing code.

muffgaga · 2024-02-27T14:20:54Z

Just a comment: On my system (macOS 14.3.1) with cmake 3.28.3, -Dwith-mpi=OFF works as advertised. With this setting, the MPI-tests are not performed at all: […]

So it seems to me that the ON/OFF flag works in a number of cases at least with the existing code.

Was this with a hdf5 dependency that was build with MPI support (I couldn't find it in your answer).
In my case, the with/without-mpi works fine as long as hdf5 hasn't been built with MPI support (I probably should have stated this more clearly).

heplesser · 2024-02-27T16:27:17Z

Was this with a hdf5 dependency that was build with MPI support (I couldn't find it in your answer). In my case, the with/without-mpi works fine as long as hdf5 hasn't been built with MPI support (I probably should have stated this more clearly).

Could it be that hdf5 was built with a different MPI version than is being used to build NEST? Could one find out which MPI was used to build hdf5, so one could pass that to -Dwith-mpi=... to ensure NEST is built against the same MPI?

muffgaga · 2024-02-28T16:52:06Z

Was this with a hdf5 dependency that was build with MPI support (I couldn't find it in your answer). In my case, the with/without-mpi works fine as long as hdf5 hasn't been built with MPI support (I probably should have stated this more clearly).

Ok, I'll update my initial #2993 to reflect that info.

Could it be that hdf5 was built with a different MPI version than is being used to build NEST? Could one find out which MPI was used to build hdf5, so one could pass that to -Dwith-mpi=... to ensure NEST is built against the same MPI?

That's exactly what the package manager spack is all about — it creates a dependency tree (a DAG) that fulfills all the requirements from the requested packages on the top (e.g. nest — but this usually includes multiple packages that one might want to use "together", e.g. elephant, nest and neo) down to all the leaves. If this isn't possible, it fails to find a solution and reports this back. Each package can be constrained in terms of version (specific or range), and/or variant… e.g. nest+mpi~sonata for a nest with mpi but without sonata support — in this particular case it depends on a mpi package (this one is virtual, because there are multiple packages that provide "mpi") but not on hdf5. The spack tool also provides environment variables (like include paths, linker paths) according to the specified dependencies… i.e. if one wants to build nest with mpi, the dependency on mpi is "resolved", a particular implementation is chosen (on HPC sites, this is usually something that is "externally" provided) and the corresponding paths are injected during the build of nest.
(That's basically the "EBRAINS Software Distribution in a nutshell" ;) )

cmake/ProcessOptions.cmake

Co-authored-by: Hans Ekkehard Plesser <hans.ekkehard.plesser@nmbu.no>

heplesser · 2024-03-05T18:37:28Z

@gtrensch Pylint failed due to a change in Pylint for which we already have a fix in master. Could you pull master in here so tests pass again?

@muffgaga Could you check if the newest version of this PR fixes your problem?

muffgaga · 2024-03-12T11:24:55Z

@heplesser I did test it with the current state and (unfortunately) see the same behavior as before — i.e. when building nest with sonata but w/o mpi against a hdf5 with mpi it fails…

nest@3.6~mpi+sonata ^hdf5~mpi ✔️
nest@3.6~mpi+sonata ^hdf5+mpi ⚔️

-- Found Boost: /opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/boost-1.79.0-cq4pl77kehytn56ey5zjajcsjzpzlekl/lib/cmake/Boost-1.79.0/BoostConfig.cmake (found suitable version "1.79.0", minimum required is "1.69.0")  
-- Checking for module 'mpi-c'
--   Package 'mpi-c', required by 'virtual:world', not found
-- Checking for module 'mpi-cxx'
--   Package 'mpi-cxx', required by 'virtual:world', not found
CMake Error at /opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindPackageHandleStandardArgs.cmake:230 (message):
  Could NOT find MPI (missing: MPI_C_FOUND MPI_CXX_FOUND)

nest@3.6+mpi+sonata ^hdf5+mpi ✔️
nest@3.6+mpi+sonata ^hdf5~mpi ✔️

I suspect it's somehow the find_package of the hdf5 dependency that pulls in the mpi check;
hdf5 installs a cmake/hdf5-config.cmake which might be related — this one might be loaded via nest's find_package( HDF5 … in cmake/ProcessOptions.cmake (sorry, no cmake expert here 😬 )?

In case of the failing one (i.e. hdf5 with mpi), it looks like this:

# …
set (${HDF5_PACKAGE_NAME}_ENABLE_PARALLEL       ON)
# …
if (${HDF5_PACKAGE_NAME}_ENABLE_PARALLEL)
  …
  find_package(MPI QUIET REQUIRED)
endif ()

In the failing case set (${HDF5_PACKAGE_NAME}_ENABLE_PARALLEL OFF) and the conditional find_package isn't taken (I guess).

heplesser · 2024-03-12T11:36:59Z

@muffgaga Thank you for the very detailed explanation. I now understand the problem much better. But it also makes me wonder why you would want to build NEST without MPI on a system that supports MPI (since HDF5 is built with MPI)?

muffgaga · 2024-03-12T12:00:46Z

@muffgaga Thank you for the very detailed explanation. I now understand the problem much better. But it also makes me wonder why you would want to build NEST without MPI on a system that supports MPI (since HDF5 is built with MPI)?

Actually, I wouldn't… I would expect that cmake (when calling it for the nest build) either throws an error or that it would override my selection and enable MPI.

cmake/ProcessOptions.cmake

heplesser · 2024-03-12T12:42:44Z

@muffgaga Thank you for the very detailed explanation. I now understand the problem much better. But it also makes me wonder why you would want to build NEST without MPI on a system that supports MPI (since HDF5 is built with MPI)?

Actually, I wouldn't… I would expect that cmake (when calling it for the nest build) either throws an error or that it would override my selection and enable MPI.

CMake throws an error, just a very confusing one, doesn't it? And I would not want CMake (or any other software) to silently switch configuration settings.

What really confuses me is that HDF5 seems to have an effect on MPI-detection, even though the HDF5-check happens several steps later than the MPI check.

Could you run CMake with --debug-output and send me the output and the tarred build directory so I can did deeper into the CMake trouble?

I would suggest to move further discussion back to #2993, as I will close this PR soon in favor of a corresponding one dedicated to just making -Dwith-mpi=path work.

heplesser · 2024-03-12T12:54:21Z

This will be replaced by #3148. Keeping open temporarily until discussion is back on #2993, but turning to draft as it should not be merged.

muffgaga · 2024-03-12T14:06:30Z

Actually, I wouldn't… I would expect that cmake (when calling it for the nest build) either throws an error or that it would override my selection and enable MPI.

CMake throws an error, just a very confusing one, doesn't it? And I would not want CMake (or any other software) to silently switch configuration settings.

Yes, that's technically correct (which is the best kind of correct), but it's a "partial MPI config check"-type of error instead of a "We don't expect MPI to be used when it's explicitly disabled by the user"… ;)

What really confuses me is that HDF5 seems to have an effect on MPI-detection, even though the HDF5-check happens several steps later than the MPI check.

Could you run CMake with --debug-output and send me the output and the tarred build directory so I can did deeper into the CMake trouble?

ACK, will do (it's running).

I would suggest to move further discussion back to #2993, as I will close this PR soon in favor of a corresponding one dedicated to just making -Dwith-mpi=path work.

👍

muffgaga · 2024-03-12T14:13:17Z

And nice, the --debug-output options shows the stack — it seems to be like I initially guessed → via the hdf5.

(I'll send the tar ball via some chat.)

--   Package 'mpi-cxx', required by 'virtual:world', not found
   Called from: [7]	/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindPkgConfig.cmake
                [6]	/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindPkgConfig.cmake
                [5]	/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindMPI.cmake
                [4]	/fasthome/mueller/code/ebrains-spack-builds/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/hdf5-1.14.3-wbpbob3l2cdywg4z2ig2xkctyff4qlyc/cmake/hdf5-config.cmake
                [3]	/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindHDF5.cmake
                [2]	/tmp/mueller/spack-stage/spack-stage-nest-3.6-6t3syywwatejmtff6phqb224vf3pabz5/spack-src/cmake/ProcessOptions.cmake
                [1]	/tmp/mueller/spack-stage/spack-stage-nest-3.6-6t3syywwatejmtff6phqb224vf3pabz5/spack-src/CMakeLists.txt
CMake Error at /opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindPackageHandleStandardArgs.cmake:230 (message):
  Could NOT find MPI (missing: MPI_C_FOUND MPI_CXX_FOUND)
Call Stack (most recent call first):
  /opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindPackageHandleStandardArgs.cmake:600 (_FPHSA_FAILURE_MESSAGE)
  /opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindMPI.cmake:1837 (find_package_handle_standard_args)
  /fasthome/mueller/code/ebrains-spack-builds/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/hdf5-1.14.3-wbpbob3l2cdywg4z2ig2xkctyff4qlyc/cmake/hdf5-config.cmake:105 (find_package)
  /opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindHDF5.cmake:486 (find_package)
  cmake/ProcessOptions.cmake:612 (find_package)
  CMakeLists.txt:154 (nest_process_with_hdf5)


   Called from: [7]	/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindPackageHandleStandardArgs.cmake
                [6]	/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindPackageHandleStandardArgs.cmake
                [5]	/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindMPI.cmake
                [4]	/fasthome/mueller/code/ebrains-spack-builds/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/hdf5-1.14.3-wbpbob3l2cdywg4z2ig2xkctyff4qlyc/cmake/hdf5-config.cmake
                [3]	/opt/spack/opt/spack/linux-debian11-sandybridge/gcc-11.2.0/cmake-3.26.3-dh7s6rreortinophvufzshblcnipxvsk/share/cmake-3.26/Modules/FindHDF5.cmake
                [2]	/tmp/mueller/spack-stage/spack-stage-nest-3.6-6t3syywwatejmtff6phqb224vf3pabz5/spack-src/cmake/ProcessOptions.cmake
                [1]	/tmp/mueller/spack-stage/spack-stage-nest-3.6-6t3syywwatejmtff6phqb224vf3pabz5/spack-src/CMakeLists.txt
-- Configuring incomplete, errors occurred!

heplesser · 2024-03-13T10:09:30Z

Closing this one now, has been superseeded by #3148.

Fix processing of CMake option with-mpi

0d565f5

gtrensch requested a review from muffgaga February 23, 2024 08:57

heplesser self-requested a review February 23, 2024 09:27

heplesser added T: Bug Wrong statements in the code or documentation S: High Should be handled next I: No breaking change Previously written code will work as before, no one should note anything changing (aside the fix) labels Feb 23, 2024

muffgaga mentioned this pull request Feb 28, 2024

failing spack build of nest@3.6 (current "ebrains version") without mpi #2993

Open

heplesser reviewed Mar 1, 2024

View reviewed changes

cmake/ProcessOptions.cmake Show resolved Hide resolved

Update cmake/ProcessOptions.cmake

3e722d7

Co-authored-by: Hans Ekkehard Plesser <hans.ekkehard.plesser@nmbu.no>

terhorstd added this to the NEST 3.7 milestone Mar 4, 2024

heplesser reviewed Mar 12, 2024

View reviewed changes

cmake/ProcessOptions.cmake Outdated Show resolved Hide resolved

Update cmake/ProcessOptions.cmake

3a5cddd

heplesser mentioned this pull request Mar 12, 2024

Make -Dwith-mpi respect provided path and ensure "default off" options are turned on only by explicit ON #3148

Merged

heplesser marked this pull request as draft March 12, 2024 12:54

heplesser removed this from the NEST 3.7 milestone Mar 12, 2024

heplesser closed this Mar 13, 2024

terhorstd deleted the fix-issue2993-with-mpi-option branch October 17, 2025 07:30

Fix processing of CMake option with-mpi #3112

Fix processing of CMake option with-mpi #3112

Uh oh!

Conversation

gtrensch commented Feb 23, 2024 • edited by heplesser Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

muffgaga commented Feb 23, 2024

Uh oh!

heplesser commented Feb 23, 2024

Uh oh!

terhorstd commented Feb 26, 2024

Uh oh!

muffgaga commented Feb 26, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

heplesser commented Feb 27, 2024

Uh oh!

muffgaga commented Feb 27, 2024

Uh oh!

heplesser commented Feb 27, 2024

Uh oh!

muffgaga commented Feb 28, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

heplesser commented Mar 5, 2024

Uh oh!

muffgaga commented Mar 12, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

heplesser commented Mar 12, 2024

Uh oh!

muffgaga commented Mar 12, 2024

Uh oh!

Uh oh!

heplesser commented Mar 12, 2024

Uh oh!

heplesser commented Mar 12, 2024

Uh oh!

muffgaga commented Mar 12, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

muffgaga commented Mar 12, 2024

Uh oh!

heplesser commented Mar 13, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Fix processing of CMake option `with-mpi` #3112

Fix processing of CMake option `with-mpi` #3112

gtrensch commented Feb 23, 2024 •

edited by heplesser

Loading

muffgaga commented Feb 26, 2024 •

edited

Loading

muffgaga commented Feb 28, 2024 •

edited

Loading

muffgaga commented Mar 12, 2024 •

edited

Loading

muffgaga commented Mar 12, 2024 •

edited

Loading