Installing with metaconfigure

These instructions are for building NJOY21 prior to version 1.2.0. More up to date instructions can be found on the install page.

Both NJOY21 and NJOY2016 use the same configuration and build process. Additionally, all of the supporting components use the same process.

For the impatient

# Download the source code
git clone https://github.com/njoy/NJOY21.git

# Get the desired version of NJOY21 (1.1.0 in this example)
cd NJOY21
wget https://raw.githubusercontent.com/njoy/signatures/master/NJOY21/1.1.0-NJOY21.json
./metaconfigure/fetch_subprojects.py 1.1.0-NJOY21.json

# Configure the build process
mkdir bin
cd bin
cmake -D fetched_subprojects=true ../

# Build NJOY1
make

# Test NJOY1
make test

Prerequisites

Development for NJOY uses the latest published language standards that are widely supported by compiler vendors, at the start of development. Therefore, the minimum language standards are

Additionally, we use CMake to configure the build system and git for version control.

Build Process

Those interested in the NJOY21 development version should read here for slightly different build instructions.

Downloading

To download NJOY21, simply git clone the repository. First move into the directory where you want the source code, then execute:

git clone https://github.com/njoy/NJOY21.git

In addition, you should download a signature file for the version of NJOY21 for which you are interested. The signatures can be found in our repository on GitHub at https://github.com/njoy/signatures/NJOY21. For this example we will use the signature file: 1.1.0-NJOY21.json. Save this file inside the NJOY21 directory that was created during the git clone operation.

wget https://raw.githubusercontent.com/njoy/signatures/master/NJOY21/1.1.0-NJOY21.json

You can inspect the json file before downloading it by looking at the signatures repository on GitHub.

Similarly for NJOY2016

git clone https://github.com/njoy/NJOY2016.git

Configuring

The configuration and build is performed in a directory (bin) inside the source directory

cd NJOY21
./metaconfigure/fetch_subprojects.py ../1.1.0-NJOY21.json
mkdir bin
cd bin
cmake -D fetched_subprojects=true ../

Note this will require a connection to the internet as cmake command will download (clone) the necessary dependencies. These will be placed in the dependencies directory.

Compiling/Building

make

You can provide the command-line option -j n (where n is the number of concurrent/parallel “jobs”) to speed up the compilation step.

Testing

make test

Like for the compiling step, you can pass the -j command-line option here as well.

If all the tests pass, you should see something that looks like:

100% tests passed, 0 tests failed out of 90

Building Offline

For this configuration/build process, a connection to the internet is required as subprojects are downloaded from GitHub. There are many instances when one would need to build NJOY on a machine that is not connected to the internet. To do that, please use these steps:

# Download the source code
git clone https://github.com/njoy/NJOY21.git

# Configure the build process
cd NJOY21
./metaconfigure/fetch_subprojects.py ../1.1.0-NJOY21.json

#### Transfer the entire NJOY21 directory to machine
#### not connected to the internet

cd NJOY21
mkdir bin
cd bin
cmake -D fetched_subprojects=TRUE ../

# Build NJOY21
make

# Test NJOY21
make test

Compliant Compilers

We have tested NJOY2016 and NJOY21 with the following compilers.

C++17 Compliant Compiler

  • Mac:

    • The clang compiler that ships with XCode will be sufficient

  • Linux:

  • Windows:

    Note that in Windows we currently support compiling and running under the Windows Subsystem for Linux or the Cygwin environment. If you have the necesssary components installed, this should also work in the DOS environment.

Fortran 2003 Compliant Compiler

  • Linux:
    • gfortran This is included as part of the standard gcc suite of compilers.
  • Mac:
  • Windows:
    • gfortran This is included as part of the standard gcc suite of compilers.

Other compilers will most likely compile without any problems, but may give different answers. The differences are typically small and due to different ways of optimizing the code from each compiler.