HDF5 Fortran Errors

[scopatz]

Hello All,

I am trying to build a conda package for OpenMC for conda-forge. I am running into many errors of the form:

[ 71%] Building Fortran object CMakeFiles/openmc.dir/src/angle_distribution.F90.o
/home/scopatz/miniconda/conda-bld/openmc_1475272242147/work/openmc-0.8.0/src/hdf5_interface.F90:2407.60:

      call h5dread_f(dset_id, dtype_real, f_ptr_r, hdf5_err)
                                                            1
Error: There is no specific subroutine for the generic 'h5dread_f' at (1)

I am not sure where this is coming from and various flags to fix this (like -fdefault-integer-8) then break fox.

You can find the recipe and the build script I am using here: https://github.com/scopatz/staged-recipes/tree/openmc/recipes/openmc

Can anyone point me inthe right direction here, please? Or maybe better yet, if a core dev of OpenMC wants to take over the recipe that would be even better! I could then help make sure it gets through the conda-forge process.

Thanks in advance!

[scopatz]

Also CC @FlanFlanagan, just so he is aware.

[nelsonag]

Any idea if that hdf5 Liv was compiled with Fortran support? (E.g., https://git.archlinux.org/svntogit/packages.git/tree/trunk/PKGBUILD?h=packages/hdf5-cpp-fortran )

On Sep 30, 2016 5:58 PM, "Anthony Scopatz" notifications@github.com wrote:

Also CC @FlanFlanagan https://github.com/FlanFlanagan, just so he is aware.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/mit-crpg/openmc/issues/727#issuecomment-250862241, or mute the thread https://github.com/notifications/unsubscribe-auth/AA_TM8eKbYt02BfFt3EtcioadVT-YeMTks5qvYYdgaJpZM4KLjHp .

[scopatz]

Yes, it was. You can see the script for how HDF5 was compiled here: https://github.com/conda-forge/hdf5-feedstock/blob/master/recipe/build.sh#L24

[paulromano]

If you're using HDF5 < 1.10, you also need to specify --enable-fortran2003.

[scopatz]

Thanks @paulromano, adding --enable-fortran2003 to the OpenMC cmake call didn't work and produces the same failures.

[scopatz]

Or do you mean to the HDF5 build?

[paulromano]

Yes, on the HDF5 build itself.

On Sep 30, 2016 6:38 PM, "Anthony Scopatz" notifications@github.com wrote:

Or do you mean to the HDF5 build?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/mit-crpg/openmc/issues/727#issuecomment-250875849, or mute the thread https://github.com/notifications/unsubscribe-auth/AAtWt4pX5D3NxmkCJInNulzODrYqNUMTks5qvZ1mgaJpZM4KLjHp .

[paulromano]

@scopatz Closing this as I am assuming --enable-fortran2003 fixed the problems. Feel free to reopen if you experience further build issues.

[scopatz]

Hi Paul, I think closing this was premature as I haven't confirmed that f03 was the issue. For whatever reason, it is taking a long time to get the new build of HDF5 approved....

[paulromano]

Premature, maybe, but given that we know the prior HDF5 installation didn't have --enable-fortran2003, one would expect to get the exact error you received. We've also had a number of users run into the same problem, so my closing the issue was based on it being a known problem with a known resolution. Again, feel free to reopen if you're still having issues after the new build of HDF5 goes through.

[scopatz]

I don't have permissions to reopen.

[paulromano]

Very well then. I'll reopen and you can close at your leisure :)

[scopatz]

Thanks! I have been working on this throughout the weekend, but not everyone does that and so I am not surprised that the update wasn't merged yet.

[scopatz]

Ok the new hdf5 package does seem to work. Thanks @nelsonag @paulromano