Installation

Dependencies

Required

• Python 3.x

• ecFlow

Optional

• Jupyter Notebook

Setup

Conda Environment (Recommended)

An environment definition file to setup Conda is provided at the root of the pyflow repository. The following commands create a suitable environment with all necessary dependencies, including ecFlow:

# clone using HTTPS
git clone https://github.com/ecmwf/pyflow.git
# or, clone using an SSH key
git clone git@github.com:ecmwf/pyflow.git

cd pyflow
conda env create -f environment.yml
conda activate pyflow

To install pyflow in the Conda environment, simply use pip

pip install pyflow-workflow-generator

Install from Source

Follow the ecFlow installation instructions ensuring that the ecFlow Python interface is enabled (see ENABLE_PYTHON). After a successful build, perform the ecFlow installation step to place all binary artifacts into an instalation directory. Considering $ECFLOW_DIR is the installation directory, the ecFlow Python interface is available at $ECFLOW_DIR/lib/python3.XX/site-packages/ecflow/.

Ensure the following environment variables are set, and install pyflow:

export ECFLOW_DIR=/path/to/installation/directory/of/ecflow

# clone using HTTPS
git clone https://github.com/ecmwf/pyflow.git
# or, clone using an SSH key
git clone git@github.com:ecmwf/pyflow.git

cd pyflow
pip install .