occwl/docs/occwl_developers_guide.md

OCCWL developers and maintainers guide

Installing locally

• Clone OCCWL into a local directory
• CD into the cloned repository: cd /path/to/occwl
• Create an environment

conda env create -f environment.yml
conda activate occwl_dev

Then you can either use setuptools to install occwl locally

pip install .

or you can add the full path to the src/occwl folder to the PYTHONPATH environment variable:

• Linux:

 export PYTHONPATH=$PYTHONPATH:/path/to/occwl/src/

• Windows:

set PYTHONPATH=%PYTHONPATH%;path\to\occwl\src

To check everything is working run

python -m unittest

Running the tests

To run the occwl tests use the following command

python -m unittest tests

Build Conda Package from Source and Upload to Anaconda

• Build and test package locally:

  1. Set the package version by modifying version parameters in ci/conda/meta.yaml and setup.py files.

  2. Build and test package:

  sh build_conda_package.sh
  

• Upload the new local package to Anaconda:

  1. Create an Anaconda account at https://anaconda.org/

  2. Login to Anaconda:

     anaconda login 
     

  3. Upload package to Anaconda channel:

     anaconda upload /Users/<user_name>/miniconda/conda-bld/noarch/occwl-<version>-py_0.tar.bz2
     

• Build and test package and upload to Anaconda:

  1. Create an Anaconda account at https://anaconda.org/

  2. Set the package version by modifying version parameters in ci/conda/meta.yaml and setup.py

  3. Build and test package and upload to Anaconda:

  sh build_conda_package.sh <anaconda_username> 
  

Install Conda Package

• Install conda package from local channel

  conda install --use-local occwl
  

  or

  conda install -c file:///Users/<user_name>/miniconda/envs/<env_name>/conda-bld/ occwl
  

• Install conda Package from Anaconda

  conda install -c <anaconda_username> -c conda-forge occwl
  

  Note that packages in Anaconda account are public by default and can be installed by other usesr without login to Anaconda.

Note the following dependencies will also be automatically installed:

• pythonocc-core
• pyqt
• pyside2
• wxpython
• numpy
• networkx
• pyDeprecate

However, in Linux platform, pyqt requires an additional system dependency, libgl1-mesa-glx, which needs to be installed as follows:

sudo apt update
sudo apt install libgl1-mesa-glx