Setup the environment¶

General checks¶

perflint physipy

Release¶

On Github¶

Once the package is shipped to pypi, you can create a release on Github : (help from github) : - Go to the package home - On the right-side menu, click on Release - Draft a new release - Select the tag version - Write a title and description - The source code from the repo are packed in zip/tar.gz files. You can add wheel or other packages by drag-n-drop.

Testing¶

Run unittests¶

python -W ignore -m unittest -v

Type hints, type checking¶

mypy ./physipy

With doctest¶

Run each file with doctest:

python -m doctest -v physipy/quantity/dimension.py

Another approach is to use pytest to run all tests, including the doctests

pytest --doctest-modules
pytest -v --doctest-modules --doctest-glob="*.py"
pytest -vv --doctest-modules --doctest-glob="*.py" .\physipy\

Note that using python -m doctest -v .\physipy\quantity\utils.py (for example) fails because of relative import, as the script is considered a stand-alone file.

Sorting imports¶

A simple good practice is to sort imports using the isort package: pip install isort isort .\physipy\ --verbose To run just check to see if anything import should be sorted : isort . --check-only

Formatting¶

pycodestyle --config=setup.cfg --verbose --statistics -qq . Given an error number : pycodestyle --show-source --select=E722 .

Benchmarking versions¶

Benchmark results using asv are available at https://mocquin.github.io/physipy/ :

For more information, see the dedicated documentation page on [benchmarking with airspeedvelocity].

For documentation¶

The documentation system of physipy is based on [mkdocs]. It relies additionnaly on [mkdocs-material] for easy customisation and [mkdocstrings] to automaticaly include all physipy's docstrings into the documentation.

pip install mkdocs
pip install mkdocs-material
pip install mkdocstrings-python
pip install markdown_include : to include the repo README.md directly into the documentation site

The documentation configuration is set in the mkdocs.yml configuration file at the root of the project.

The raw documentation is stored in the [docs] directory as markdown files. Markdown files can converted back-and-forth with notebook format, using :

jupyter nbconvert --to markdown .\docs\scientific-stack\math-support.ipynb # from nb to md
jupytext --to ipynb .\docs\scientific-stack\numpy-support.md               # from md to nb

During development, cd to physipy root directory, and run the following command to launch a live server : mkdocs serve

Please note that this only serves the documentation, and does not export it to an actual static-website. To export the documentation to a static-website use : mkdocs build

This will generate the content to a directory given by site_dir in the mkdocs.yml configuration file (currently in _mkdocs_site).

To pin the packages needed for the doc, create a new venv, install pip-tools, then the packages specified in docs/requirements.in. Use mkdocs serve to review the docs, and then pip-compile docs/requirements.in to create a docs/requirements.txt file.