RDFLib developers guide

Introduction

This document describes the process and conventions to follow when developing RDFLib code.

  • Please be as Pythonic as possible (PEP 8).

  • Code should be formatted using black and we use Black v23.1.0, with the black config in pyproject.toml.

  • Code should also pass flake8 linting and mypy type checking.

  • You must supply tests for new code.

  • RDFLib uses Poetry for dependency management and packaging.

If you add a new cool feature, consider also adding an example in ./examples.

Pull Requests Guidelines

Contributions to RDFLib are made through pull requests (PRs).

For changes that add features or affect the public API of RDFLib, it is recommended to first open an issue to discuss the change before starting to work on it. That way you can get feedback on the design of the feature before spending time on it.

In general, maintainers will only merge PRs if the following conditions are met:

  • The PR has been sufficiently reviewed.

    Each PR should be reviewed and approved by at least two people other than the author of the PR before it is merged and PRs will be processed faster if they are easier to review and approve of.

    Reviews are open to everyone, but the weight assigned to any particular review is at the discretion of maintainers.

  • Changes that have a runtime impact are covered by unit tests.

    There should either be existing tests that cover the changed code and behaviour, or the PR should include tests. For more information about what is considered adequate testing see the Tests section.

  • Documentation that covers something that changed has been updated.

  • Type checks and unit tests that are part of our continuous integration workflow pass.

In addition to these conditions, PRs that are easier to review and approve will be processed quicker. The primary factors that determine this are the scope and size of a PR. If there are few changes and the scope is limited, then there is less that a reviewer has to understand and less that they can disagree with. It is thus important to try to split up your changes into multiple independent PRs if possible. No PR is too small.

For PRs that introduce breaking changes, it is even more critical that they are limited in size and scope, as they will likely have to be kept up to date with the main branch of this project for some time before they are merged.

It is also critical that your PR is understandable both in what it does and why it does it, and how the change will impact the users of this project, for this reason, it is essential that your PR’s description explains the nature of the PR, what the PR intends to do, why this is desirable, and how this will affect the users of this project.

Please note that while we would like all PRs to follow the guidelines given here, we will not reject a PR just because it does not.

Maintenance Guidelines

This section contains guidelines for maintaining RDFLib. RDFLib maintainers should try to follow these. These guidelines also serve as an indication to RDFLib users what they can expect.

Breaking changes

Breaking changes to RDFLib’s public API should be made incrementally, with small pull requests to the main branch that change as few things as possible.

Breaking changes should be discussed first in an issue before work is started, as it is possible that the change is not necessary or that there is a better way to achieve the same goal, in which case the work on the PR would have been wasted. This will however not be strictly enforced, and no PR will be rejected solely on the basis that it was not discussed upfront.

RDFLib follows semantic versioning and trunk-based development, so if any breaking changes were introduced into the main branch since the last release, then the next release will be a major release with an incremented major version.

Releases of RDFLib will not as a rule be conditioned on specific features, so there may be new major releases that contain very few breaking changes, and there could be no minor or patch releases between two major releases.

Rationale

RDFLib has been around for more than a decade, and in this time both Python and RDF have evolved, and RDFLib’s API also has to evolve to keep up with these changes and to make it easier for users to use. This will inevitably require breaking changes.

There are more or less two ways to introduce breaking changes to RDFLib’s public API:

  • Revolutionary: Create a new API from scratch and reimplement it, and when ready, release a new version of RDFLib with the new API.

  • Evolutionary: Incrementally improve the existing API with small changes and release any breaking changes that were made at regular intervals.

While the revolutionary approach seems appealing, it is also risky and time-consuming.

The evolutionary approach puts a lot of strain on the users of RDFLib as they have to adapt to breaking changes more often, but the shortcomings of the RDFLib public API also put a lot of strain on the users of RDFLib. On the other hand, a major advantage of the evolutionary approach is that it is simple and achievable from a maintenance and contributor perspective.

Deprecating functionality

To whatever extent possible, classes, functions, variables, or parameters that will be removed should be marked for deprecation in documentation, and if possible, should be changed to raise deprecation warnings if used.

There is however no hard requirement that something may only be removed after a deprecation notice has been added, or only after a release was made with a deprecation notice.

Consequently, functionality may be removed without it ever being marked as deprecated.

Rationale

Current resource limitations and the backlog of issues make it impractical to first release or incorporate deprecation notices before making quality of life changes.

RDFLib uses semantic versioning and provides type hints, and these are the primary mechanisms for signalling breaking changes to our users.

Tests

Any new functionality being added to RDFLib must have unit tests and should have doc tests supplied.

Typically, you should add your functionality and new tests to a branch of RDFlib and run all tests locally and see them pass. There are currently close to 4,000 tests, with a some expected failures and skipped tests. We won’t merge pull requests unless the test suite completes successfully.

Tests that you add should show how your new feature or bug fix is doing what you say it is doing: if you remove your enhancement, your new tests should fail!

Finally, please consider adding simple and more complex tests. It’s good to see the basic functionality of your feature tests and then also any tricky bits or edge cases.

Testing framework

RDFLib uses the pytest testing framework.

Running tests

To run RDFLib’s test suite with pytest:

$ poetry install
$ poetry run pytest

Specific tests can be run by file name. For example:

$ poetry run pytest test/test_graph/test_graph.py

For more extensive tests, including tests for the berkleydb backend, install extra requirements before executing the tests.

$ poetry install --all-extras
$ poetry run pytest

Writing tests

New tests should be written for pytest instead of for python’s built-in unittest module as pytest provides advanced features such as parameterization and more flexibility in writing expected failure tests than unittest.

A primer on how to write tests for pytest can be found here.

The existing tests that use unittest work well with pytest, but they should ideally be updated to the pytest test-style when they are touched.

Test should go into the test/ directory, either into an existing test file with a name that is applicable to the test being written, or into a new test file with a name that is descriptive of the tests placed in it. Test files should be named test_*.py so that pytest can discover them.

Running static checks

Check formatting with black, making sure you use our black.toml config file:

poetry run black .

Check style and conventions with flake8:

poetry run flake8 rdflib

We also provide a flakeheaven baseline that ignores existing flake8 errors and only reports on newly introduced flake8 errors:

poetry run flakeheaven

Check types with mypy:

poetry run mypy --show-error-context --show-error-codes

pre-commit and pre-commit ci

We have pre-commit configured with black for formatting code.

Some useful commands for using pre-commit:

# Install pre-commit.
pip install --user --upgrade pre-commit

# Install pre-commit hooks, this will run pre-commit
# every time you make a git commit.
pre-commit install

# Run pre-commit on changed files.
pre-commit run

# Run pre-commit on all files.
pre-commit run --all-files

There is also two tox environments for pre-commit:

# run pre-commit on changed files.
tox -e precommit

# run pre-commit on all files.
tox -e precommitall

There is no hard requirement for pull requests to be processed with pre-commit (or the underlying processors), however doing this makes for a less noisy codebase with cleaner history.

We have enabled https://pre-commit.ci/ and this can be used to automatically fix pull requests by commenting pre-commit.ci autofix on a pull request.

Using tox

RDFLib has a tox config file that makes it easier to run validation on all supported python versions.

# Install tox.
pip install tox

# List the tox environments that run by default.
tox -e

# Run the default environments.
tox

# List all tox environments, including ones that don't run by default.
tox -a

# Run a specific environment.
tox -e py38 # default environment with py37
tox -e py39-extra # extra tests with py39

# Override the test command.
# the below command will run `pytest test/test_translate_algebra.py`
# instead of the default pytest command.
tox -e py38,py39 -- pytest test/test_translate_algebra.py

go-task and Taskfile.yml

A Taskfile.yml is provided for go-task with various commands that facilitate development.

Instructions for installing go-task can be seen in the go-task installation guide.

Some useful commands for working with the task in the taskfile is given below:

# List available tasks.
task -l

# Configure the environment for development
task configure

# Run basic validation
task validate

# Build docs
task docs:build

# Run live-preview on the docs
task docs:live-server

# Run the py310 tox environment
task tox -- -e py310

The Taskfile usage documentation provides more information on how to work with taskfiles.

Development container

To simplify the process of getting a working development environment to develop rdflib in we provide a Development Container (devcontainer) that is configured in Docker Compose. This container can be used directly to run various commands, or it can be used with editors that support Development Containers.

Important

The devcontainer is intended to run with a rootless docker daemon so it can edit files owned by the invoking user without an invovled configuration process.

Using a rootless docker daemon also has general security benefits.

To use the development container directly:

# Build the devcontainer docker image.
docker-compose build

# Configure the system for development.
docker-compose run --rm run task configure

# Run the validate task inside the devtools container.
docker-compose run --rm run task validate

# Run extensive tests inside the devtools container.
docker-compose run --rm run task EXTENSIVE=true test

# To get a shell into the devcontainer docker image.
docker-compose run --rm run bash

The devcontainer also works with Podman Compose.

Details on how to use the development container with VSCode can found in the Developing inside a Container page. With the VSCode development container CLI installed the following command can be used to open the repository inside the development container:

# Inside the repository base directory
cd ./rdflib/

# Build the development container.
devcontainer build .

# Open the code inside the development container.
devcontainer open .

Writing documentation

We use sphinx for generating HTML docs, see Writing RDFLib Documentation.

Continuous Integration

We used GitHub Actions for CI, see:

If you make a pull-request to RDFLib on GitHub, GitHub Actions will automatically test your code and we will only merge code passing all tests.

Please do not commit tests you know will fail, even if you’re just pointing out a bug. If you commit such tests, flag them as expecting to fail.

Compatibility

RDFlib 7.0.0 release and later only support Python 3.8.1 and newer.

RDFlib 6.0.0 release and later only support Python 3.7 and newer.

RDFLib 5.0.0 maintained compatibility with Python versions 2.7, 3.4, 3.5, 3.6, 3.7.

Releasing

Create a release-preparation pull request with the following changes:

  • Updated version and date in CITATION.cff.

  • Updated copyright year in the LICENSE file.

  • Updated copyright year in the docs/conf.py file.

  • Updated main branch version and current version in the README.md file. The main branch version should be the next major version with an a0 suffix to indicate it is alpha 0. When releasing 6.3.1, the main branch version in the README should be 6.4.0a0.

  • Updated version in the pyproject.toml file.

  • Updated __date__ in the rdflib/__init__.py file.

  • Accurate CHANGELOG.md entry for the release.

Once the PR is merged, switch to the main branch, build the release and upload it to PyPI:

# Clean up any previous builds
\rm -vf dist/*

# Build artifacts
poetry build

# Verify package metadata
bsdtar -xvf dist/rdflib-*.whl -O '*/METADATA' | view -
bsdtar -xvf dist/rdflib-*.tar.gz -O '*/PKG-INFO' | view -

# Check that the built wheel and sdist works correctly:
pipx run --no-cache --spec "$(readlink -f dist/rdflib*.whl)" rdfpipe --version
pipx run --no-cache --spec "$(readlink -f dist/rdflib*.whl)" rdfpipe https://github.com/RDFLib/rdflib/raw/main/test/data/defined_namespaces/rdfs.ttl
pipx run --no-cache --spec "$(readlink -f dist/rdflib*.tar.gz)" rdfpipe --version
pipx run --no-cache --spec "$(readlink -f dist/rdflib*.tar.gz)" rdfpipe https://github.com/RDFLib/rdflib/raw/main/test/data/defined_namespaces/rdfs.ttl

# Dry run publishing
poetry publish --repository=testpypi --dry-run
poetry publish --dry-run

# Publish to TestPyPI
poetry publish --repository=testpypi

# Publish to PyPI
poetry publish

Once this is done, create a release tag from GitHub releases. For a release of version 6.3.1 the tag should be 6.3.1 (without a “v” prefix), and the release title should be “RDFLib 6.3.1”. The release notes for the latest version be added to the release description. The artifacts built with poetry build should be uploaded to the release as release artifacts.

The resulting release will be available at https://github.com/RDFLib/rdflib/releases/tag/6.3.1

Once this is done, announce the release at the following locations:

  • Twitter: Just make a tweet from your own account linking to the latest release.

  • RDFLib mailing list.

  • RDFLib Gitter / matrix.org chat room.

Once this is all done, create another post-release pull request with the following changes:

  • Set the just released version in docker/latest/requirements.in and run task docker:prepare to update the docker/latest/requirements.txt file.

  • Set the version in the pyproject.toml file to the next minor release with a a0 suffix to indicate alpha 0.