Non-conda installation instructions

If you’re unable to use the conda-based installation instructions, the framework can use pre-existing languages and programs on your systems without the use of conda environments for dependency management. If you choose this option, we assume you know what you’re doing: because this mode of operation is dependent on the details of each user’s system, we can only support it at a secondary priority.

Requirements

The following software needs to be available on your $PATH when the framework is run:

  • Python version 2.7: the framework will attempt to create virtualenvs for each POD.

  • NCO utilities version 4.7.6.

  • ghostscript.

  • NCL, version 6.5.0 or newer.

  • R, for the SM_ET_coupling POD only.

Configuration instructions

Configuring this mode of operation requires adding additional settings to the src/default_tests.jsonc file. This is a template/example of an input file you can use to define configuration options instead of re-typing them on the command line every time you run the framework. In addition to the settings described in Configuring framework paths, you will also need to:

  • Change the value for environment_manager from "Conda" to "Virtualenv".

  • Any values for conda_root and conda_env_root will be ignored.

  • The framework will use pip to install required python modules in new virtualenvs, which will be installed in the default location for your system’s python. To put the files in a different location, create a new setting "venv_root": <Path to virtualenv directory>.

  • Likewise, to install packages needed by R in a location other than your system default, create a new setting "r_lib_root": <path to R package directory>.

Known issues with standalone NCL installation

Many Linux distributions (Ubuntu, Mint, etc.) have offered a way of installing NCL through their system package manager (apt, yum, etc.) This method of installation is not recommended: users may encounter errors when running the example PODs provided by NCAR, even if the environment variables and search path have been added.

The recommended method to install standalone NCL is by downloading the pre-compiled binaries from https://www.ncl.ucar.edu/Download/install_from_binary.shtml. Choose a download option according to the Linux distribution and hardware, unzip the file (results in 3 folders: bin, include, lib), create a folder ncl under the directory /usr/local (requires permission) and move the 3 unzipped folders into /usr/local/ncl. Then add the following lines to the .bashrc script (under the user’s home directory; may be different if using shells other than bash, e.g., .cshrc for csh):

export NCARG_ROOT=/usr/local/ncl
export PATH:$NCARG_ROOT/bin:$PATH