Uintah plugin missing from Ubuntu 22, x86_64, binary

[tharman-git]

Greetings,

The Uintah plugin is missing from the binary: visit3_4_1.linux-x86_64-ubuntu22.tar.gz, however it is present in the visit3_4_1.linux-x86_64-debian11.tar.gz binary. Is there a compilation issue with Uintah?
Thanks,
Todd Harman
University of Utah

[biagas]

Yes there was a problem building, but I didn't keep the log files of the error. I will run the docker container again and report back the error message(s).

[tharman-git]

I successfully compiled Uintah-2.6.1 on Ubuntu 22.04.4 LTS machine with:

gcc & g++ = (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0

MPI = (Open MPI) 4.1.2

and the configure line:

#!/bin/bash

../src/configure \
   --enable-minimal \
   --enable-optimize \
   --with-fortran=no\
   --with-petsc=no\
   --with-hypre=no \
   --with-lapack=no\
   --with-blas=no \
   --with-mpi='built-in' \
  CXX=mpicxx \
  CC=mpicc \
  CFLAGS="-std=c++11 -m64 -fPIC  -O2" \
  CXXFLAGS="-std=c++11 -m64 -fPIC  -O2" \

The key bit is "-std=c++11" since there is some exception code that is not c++17 compiliant.

ISO C++17 does not allow dynamic exception specifications
  244 |   throw(ConsecutiveRangeSetException)

How would you like to proceed? The problem is clearly on my end.

[tharman-git]

If it's not too much work please update the current tarball. Thank you!

[biagas]

I uploaded a new ubuntu22 distro to our release assets page.
I appended -with-uintah-beta to the name.
https://github.com/visit-dav/visit/releases
Let me know how it works.

[markcmiller86]

@biagas that new ubuntu build also has MOAB in it and it is working! Yeah! 🎉

@iulian787 @vijaysm and I tried it on ANL's unbuntu systems and its working great.

We did discover an issue with the parallel stuff though. There is a bunch of stuff related to mpi, mpirun (which is a symlink to mpiexec.hydra and the permissions on mpiexec.hydra were missing execute permissions.

I ran into something similar with my mac and had to adjust the CMake logic that installed MPI (which I believe I committed to 3.4RC). Can you confirm that permissions for those mpi tools look correct in the asset tarball?

[markcmiller86]

@biagas the particular change I had to make to fix this permissions issue was this one...

35a038d

[biagas]

@markcmiller86 thank you for that information. Looks like I have been using the source tarball from our releases page which was generated before your merge.