Intro to development
The libraries in Scikit-HEP try to follow best practices in the community for development and deployment (though some packages in Scikit-HEP are still being converted to use best practices). The following outlines the basics for setting up a development environment. It is recommended as a basis for
.github/CONTRIBUTING.md in the packages.
CONTRIBUTING.md template (click to expand)
See the [Scikit-HEP Developer introduction][skhep-dev-intro] for a detailed description of best practices for developing Scikit-HEP packages. [skhep-dev-intro]: https://scikit-hep.org/developer/intro # Setting up a development environment You can set up a development environment by running: ```bash python3 -m venv .env source ./.env/bin/activate pip install -v -e .[dev] ``` # Post setup You should prepare pre-commit, which will help you by checking that commits pass required checks: ```bash pip install pre-commit # or brew install pre-commit on macOS pre-commit install # Will install a pre-commit hook into the git repo ``` You can also/alternatively run `pre-commit run` (changes only) or `pre-commit run --all-files` to check even without installing the hook. # Testing Use pytest to run the unit checks: ```bash pytest ```
If you want to work on Python software, you should always have a virtual environment. A user may not always have one, but a developer always should. You do not want to risk breaking your main system environment, you want full control over versions of libraries, and you want to avoid “leaking” your main environment in, causing you to not notice when you have extra dependencies. Virtual environments are disposable, while it is very hard to cleanup or update a main system environment.
Any common modern system to create environments should be fine. Here is the most basic one,
venv, that comes by default with Python 3:
python3 -m venv .env
This creates a new virtual environment in a local folder, named
.env. There are a few options, but usually they are not necessary. If you don’t mind a (very common) dependency, you can use the
virtualenv package, which has the same syntax, is a little faster, and works in Python 2 as well.
To activate the virtual environment, type:
. is short for
source, which runs the script
activate in your current shell. If you like a different shell, like fish, there are several activate scripts; the default one expects a bash-like shell. You need to run this command any time you want to use the development environment. The activation script installs a function
deactivate; type that at any time to leave the environment (or just close your shell). It also adds a bit of text to your prompt so you don’t forget that you are in an environment.
Finally, you need to install the package. Most packages support several extra options when installing; for development, you may want
[complete] (packages use different conventions, check). Here is an example:
pip install -e .[dev]
-e installs the package in “editable” mode, meaning the files are not copied to your site-packages folder, so you can edit and work with the package locally. You need to rerun
pip install -e . if there are binary components and you edit those.
Never edit your
PYTHONPATH manually, or depend on the current directory for library development.
Always use pip to install, do not call
python setup.pydirectly for building or installing. Pip inserts shims into setup.py to fix common issues and enable features like PEP 517/518 builds. The only time you directly call setup.py is when making sdists, and even then there are hacks involved.
Remember a “normal” user will always use pip. You need to simulate that in development.
You can also develop in Conda. For some packages, such as those that work with ROOT, you need to be using Conda, and it is a great way to have control over the version of Python you are using. If so, then the creation of an environment looks like this:
conda create -n env_name python=3.8
You can use
-n name or
-p path to specify the environment by name or location. The following assume you used a name, but just replace names with paths if you choose a path.
Some packages provide an environment file, either for CI or developer use. If they do, you can use
conda env create -f filename.yml to create (or
update to update) the provided environment. You can override name or location as above. And if the file is called
environment.yml, you can leave off the
-f filename entirely.
To activate an environment:
conda activate env_name
To deactivate, use
conda deactivate, or leave your shell.
Building binary components for conda packages takes some care. In general, if there are any binary components, you must use conda-forge’s build system to provide binaries to users.
IDE’s can provide useful additions, such as extended type checking, type aware completions, and more. We currently do not provide IDE files in our repositories, but PyCharm community edition is suggested by some of our members for developers looking for an IDE. It includes an extension, IdeaVIM, for VIM emulation for users used to that editor. Setting up an IDE takes extra time but often provides tools (like smart renaming) that are useful, and if you use type hints, will probably pay for the setup time quite quickly when developing.
You can instruct PyCharm to use the virtual environment (regular or Conda) that you set up, or it can follow the environment.yml or requirements to make one for you.