Solution to issue cannot be found in the documentation.
[X] I checked the documentation.
Issue
openmpi 4.1.3 is built with gfortran 7, while openmpi 4.1.2 is built with gfortran 9 (change here. This results in a runtime dep on libgfortran4 in the newer build instead of libgfortran5, meaning it can't be installed with other packages built with the current default fortran compiler, like hdf5 1.12.
I'm not sure if there's an easy way in conda_build_config.yaml to specify the old cuda compiler with the new fortran compiler. Perhaps the simplest solution is to remove the pin and just let the compiler matrix do its thing. This is a bit wasteful if we ignore run exports from the cuda compiler, but it should result in correct (if redundant) outputs.
Installed packages
mamba create --dry-run -n test hdf5=1.12.1 openmpi=4.1.3 h5py=3.6
(conflict is on libgfortran5 in hdf5 and libgfortran4 in openmpi)
Environment info
active environment : base
active env location : /opt/conda
shell level : 1
user config file : /root/.condarc
populated config files : /opt/conda/.condarc
conda version : 4.12.0
conda-build version : not installed
python version : 3.9.10.final.0
virtual packages : __linux=5.10.104=0
__glibc=2.31=0
__unix=0=0
__archspec=1=x86_64
base environment : /opt/conda (writable)
conda av data dir : /opt/conda/etc/conda
conda av metadata url : None
channel URLs : https://conda.anaconda.org/conda-forge/linux-64
https://conda.anaconda.org/conda-forge/noarch
package cache : /opt/conda/pkgs
/root/.conda/pkgs
envs directories : /opt/conda/envs
/root/.conda/envs
platform : linux-64
user-agent : conda/4.12.0 requests/2.27.1 CPython/3.9.10 Linux/5.10.104-linuxkit ubuntu/20.04.4 glibc/2.31
UID:GID : 0:0
netrc file : None
offline mode : False
Solution to issue cannot be found in the documentation.
Issue
openmpi 4.1.3 is built with gfortran 7, while openmpi 4.1.2 is built with gfortran 9 (change here. This results in a runtime dep on libgfortran4 in the newer build instead of libgfortran5, meaning it can't be installed with other packages built with the current default fortran compiler, like hdf5 1.12.
It's pinning the cuda compiler version to 10.2 here that is the cause, due to zip_keys here in conda_forge_pinning.
I'm not sure if there's an easy way in conda_build_config.yaml to specify the old cuda compiler with the new fortran compiler. Perhaps the simplest solution is to remove the pin and just let the compiler matrix do its thing. This is a bit wasteful if we ignore run exports from the cuda compiler, but it should result in correct (if redundant) outputs.
Installed packages
Environment info