New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Support for OpenMC 0.13.0/HDF5 1.12.1 #1443
Comments
Hi, and welcome to PyNE! 👋 Thanks for opening your first issue. We recommend that you include information such as the version of PyNE you're working with (eg, develop branch or a specific version), the platform you are operating on, the expected behavior, and the actual behavior you are bringing our attention to. The more deatil you provide, the better others in this community will be able to help you. |
Just confirming this behavior conda create --name openmc_with_pyne_test python=3.8
conda install -c conda-forge openmc=0.13.0 pyne
>>> UnsatisfiableError: The following specifications were found to be incompatible with each other: Mamba install gives some hints regarding moab
The feedstock for pyne is here but there are not many version limits I can also confirm that even when installing without DAGMC in the OpenMC conda install there is the same conflict, this command also fails in the same way mamba install -c conda-forge openmc=0.13.0=nodagmc* pyne Lastly I can also confirm that the conda install of openmc v0.13.0 build 0 also doesn't install in a pyne environment. This is somewhat reassuring for me as it means this was a problem before I updated the openmc conda install a few days ago
|
This appears to be an incompatibility at the GFortran layer. PyNE requires an older version of libgfortran and HDF 1.12 is built on conda-forge only with newer versions. |
I see. If HDF5 1.12 is built with a much newer version of gfortran I can see why that would cause incompatibility issues. Is the PyNE dependency on this older version of libgfortran only related to its HDF5 version, or are there other components of the code that rely on the older version of libgfortran? |
The PyNE dependency on FORTRAN is not related to HDF5. We might be able to build a conda package that is independent of FORTRAN. |
Oh is this related to ahot perhaps? |
@gonuke any way I can help with this? I'd imagine this means modifying the feedstock in pyne-feedstock and adding some CMake options... It looks like there may already be some machinery to do this in the
|
Help is always welcome!!! Turning off the spatial solvers is one part of this that already has a CMake option as you have found, but may not be the whole story. There is also some FORTRAN in the ENSDF handlers and there is not yet an option to turn that off. Furthermore, I'm not entirely sure what the consequences of turning that off would be. I'd be happy to have you try, however. |
I've been trying to build PyNE from source with the |
Okay, I've gotten PyNE to build with OpenMC v0.13.0. Full gist of terminal output is here.
I'm running into an error when I'm running the test suite:
I tried rebuilding using HDF5 1.10.6 and ran into the same error, so I don't think it's related to HDF5. I've been able to track down the error to the creation of a Looks like to go farther into this I'll need to build python from source. There are two ways I'm thinking about doing this.
Either way, I'll need to somehow get that custom build of python into the conda environment (example how to do this is here). If there is an easier way around all this or someone knows the answer I'd love to hear it. |
I have installed Python from source sucessfully using the following steps:
All tests passed sucessfully. Now that I know this works, I'll need to get this into a Conda package. Update: I've gotten the debug build of python into it's own package. I've added functionality to do this automatically in ubuntu-nuclear-software-installer. Now I need to setup the Cython debugger. Will work on this tomorrow. |
Okay, I'm still going to try to set up the Cython debugger becase I think it could help demystify what's going on, but I was able to get a
@gonuke See anything of note? |
@gonuke any ideas? |
We now have a successful build of pyne with newer OpenMC/HDF5 (#1470), so we should consider a release and update of our conda-forge image. |
Is your feature request related to a problem? Please describe.
The newest version of OpenMC, v0.13.0, was recently released. However, I have been unable to get PyNE to install alongside this version. Running
conda install -c conda-forge openmc-env
in an environment with PyNE v0.7.5 gives the following error:I have tried specifying the
nomoab_noopenmc
builds and run into the same problem.I think this is happening due to all currently available builds of OpenMC v0.13.0 requiring HDF5 1.12.1 or higher. To my knowlege, all builds of PyNE 0.7.5 are pinned to HDF5 1.10.6.
Describe the solution or outcome you'd like.
Would it be possible to relax the HDF5 version requirement in PyNE? I'm not entirely sure what parts of HDF5 PyNE uses, so it's hard for me to say whether this is an easy thing to do in practice. I've looked around the PyNE repo for version files but haven't found any, so I wouldn't know where to start. I'd love to help though if this is a small change.
Describe alternatives you've considered.
The only alternative I can think of right now to get OpenMC v0.13.0 working with PyNE with is to build both OpenMC and PyNE from source on the same HDF5 version.
The text was updated successfully, but these errors were encountered: