Initial Server Setup

Note

BWRC users: Everything described in this page is already configured in /tools/C/bag, so you can jump ahead to the next page.

BAG3++ requires multiple Python and C++ dependences. These instructions will install the dependencies on your machine.

1. Install (on CentOS or Red Hat versions >=7):

   • httpd24-curl

   • httpd24-libcurl

   • devtoolset-8 (compilers)

   • rh-git218

2. Copy the environment.yml from this link in the documentation repo, and update the prefix in the last line to your desired location. Then build a miniconda3 environment from the yml file:

   $ conda env create -f environment.yml
   

   Successful building should give all python dependencies, as well as the C++ fmt and spdlog packages.

3. Create a directory to install programs in (referred to as /path/to/programs). For all following program install steps, use the gcc that comes with devtoolset-8.

4. Download and extract cmake 3.17.0, then build:

   $ wget https://github.com/Kitware/CMake/releases/download/v3.17.0/cmake-3.17.0.tar.gz
   $ tar -xvf cmake-3.17.0.tar.gz
   $ cd cmake-3.17.0
   $ ./bootstrap --prefix=/path/to/conda/env/envname --parallel=4
   $ make -j4
   $ make install
   

5. Install magic_enum as follows:

   $ git clone https://github.com/Neargye/magic_enum.git
   $ cd magic_enum
   $ cmake -H. -Bbuild -DCMAKE_BUILD_TYPE=Release -DMAGIC_ENUM_OPT_BUILD_EXAMPLES=FALSE -DMAGIC_ENUM_OPT_BUILD_TESTS=FALSE -DCMAKE_INSTALL_PREFIX=/path/to/conda/env/envname
   $ cmake --build build
   $ cd build
   $ make install
   

6. Install yaml-cpp:

   $ git clone https://github.com/jbeder/yaml-cpp.git
   $ cd yaml-cpp
   $ cmake -B_build -H. -DCMAKE_POSITION_INDEPENDENT_CODE=ON -DCMAKE_INSTALL_PREFIX=/path/to/conda/env/envname
   $ cmake --build _build --target install -- -j 4
   

7. Install libfyaml:

   $ git clone https://github.com/pantoniou/libfyaml.git
   $ cd libfyaml
   $ ./bootstrap.sh
   $ ./configure --prefix=/path/to/conda/env/envname
   $ make -j12
   $ make install
   

8. Download HDF5 1.10 (h5py-2.10 does not work with 1.12 yet), then install:

   $ wget https://support.hdfgroup.org/ftp/HDF5/releases/hdf5-1.10/hdf5-1.10.6/src/hdf5-1.10.6.tar.gz
   $ tar -xvf hdf5-1.10.6.tar.gz
   $ cd hdf5-1.10.6
   $ ./configure --prefix=/path/to/conda/env/envname
   $ make -j24
   $ make install
   

9. Boost - download source, unzip. In directory, run:

   $ wget https://boostorg.jfrog.io/artifactory/main/release/1.72.0/source/boost_1_72_0.tar.gz
   $ tar -xvf boost_1_72_0.tar.gz
   $ cd boost_1_72_0
   $ ./bootstrap.sh --prefix=/path/to/conda/env/envname
   

10. In the resulting project-config.jam file, change the using python line to:

    using python : 3.7 : /path/to/conda/env/envname : /path/to/conda/env/envname/include/python3.7m ;
    

    If it exists, delete the line:

    path-constant ICU_PATH : /usr ;
    

11. Run:

    $ ./b2 --build-dir=_build cxxflags=-fPIC -j8 -target=shared,static --with-filesystem --with-serialization --with-program_options install | tee install.log
    

    Remember to check install.log to see if there’s any error messages (like python build error, etc.).

Note

Some users report that source .bashrc after installing all programs will result in the following error:

$ rpm: /path/anaconda3/envs/bag_py3d7_c/lib/liblzma.so.5: version `XZ_5.1.2alpha` not found (required by /lib64/librpmio.so.3)

To fix this, remove /path/to/conda/envs/bag_py3d7_c/lib/liblzma.so.5.2.5. rpm will then function correctly, assuming the rpm used is in /usr/bin/rpm.