First read the overall Qiskit project contribution guidelines. These are all included in the Qiskit documentation:
https://github.com/Qiskit/qiskit/blob/main/CONTRIBUTING.md
While it's not all directly applicable since most of it is about the Qiskit project itself and rustworkx is an independent library developed in tandem with Qiskit; the general guidelines and advice still apply here.
In addition to the general guidelines there are specific details for contributing to rustworkx, these are documented below.
Rustworkx is implemented primarily in Rust with a thin layer of Python.
Because of that, most of your code changes will involve modifications to
Rust files in src. To understand which files you need to change, we invite
you for an overview of our simplified source tree:
├── src/
│ ├── lib.rs
│ ├── tiny.rs
│ ├── large/
│ │ ├── mod.rs
│ │ ├── pure_rust_code.rs
│ │ └── more_pure_rust_code.rs
To add new functions, you will need to export them in lib.rs. lib.rs will
import functions defined in Rust modules (see the next section), and export
them to Python using m.add_wrapped(wrap_pyfunction!(your_new_function))?;
To add and change functions, you will need to modify module files. Modules contain pyfunctions
that will be exported, and can be defined either as a single file such as tiny.rs or as a
directory with mod.rs such as large/.
Rust functions that are exported to Python are annotated with #[pyfunction]. The
annotation gives them power to interact both with the Python interpreter and pure
Rust code. To change an existing function, search for its name and edit the code that
already exists.
If you want to add a new function, find the module you'd like to insert it in
or create a new one like your_module.rs. Then, start with the boilerplate bellow:
/// Docstring containing description of the function
#[pyfunction]
#[pyo3(text_signature = "(graph, /)")]
pub fn your_new_function(
py: Python,
graph: &graph::PyGraph,
) -> PyResult<()> {
/* Your code goes here */
}NOTE: If you create a new
your_module.rs, remember to declare and import it inlib.rs:mod your_module; use your_module::*;
Sometimes you will find that it is hard to organize a module in a tiny
file like tiny.rs. In those cases, we suggest moving the files to a directory
and splitting them following the structure of large/.
Module directories have a mod.rs file containing the pyfunctions. The pyfunctions
in that file then delegate most of logic by importing and calling pure Rust code from
pure_rust_code.rs and more_pure_rust_code.rs.
NOTE: Do you still have questions about making your contribution? Contact us at the #rustworkx channel in Qiskit Slack
If you're working on writing a pure rust function and it can be made generic
such that it works for any petgraph graph (if applicable) and that it has
no dependency on Python or pyo3, it probably makes sense in rustworkx-core.
rustworkx-core is a standalone rust library that's used to provide a Rust API
to both rustworkx and other rust applications or libraries. Unlike rustworkx
it's a Rust library and not a Python library and is designed to be an add-on
library on top of petgraph that provides additional graph algorithms and
functionality.
When contributing to rustworkx-core the key differences to keep in mind are that the public rust interface needs to be treated as a stable interface, which is different from rustworkx where the stable rust interface compatibility doesn't matter only the exported Python API. Additionally documentation and testing should be done via cargo doc and cargo test. It is expected that any new functionality or changes to rustworkx-core is also being used by rustworkx so test coverage is needed both via python in the rustworkx tests and via the rustworkx-core rust interface.
Once you've made a code change, it is important to verify that your change does not break any existing tests and that any new tests that you've added also run successfully. Before you open a new pull request for your change, you'll want to run the test suite locally.
The easiest way to run the test suite is to use
Nox. You can install Nox
with pip: pip install -U nox. Nox provides several advantages, but the
biggest one is that it builds an isolated virtualenv for running tests. This
means it does not pollute your system python when running. However, by default
Nox will recompile rustworkx from source every time it is run even if there
are no changes made to the rust code. To avoid this you can use the
--no-install package if you'd like to rerun tests without recompiling.
Note, you only want to use this flag if you recently ran Nox and there are no
rust code (or packaged python code) changes to the repo since then. Otherwise
the rustworkx package Nox installs in it's virtualenv will be out of date (or
missing).
Note, if you run tests outside of Nox that you can not run the tests from the root of the repo, this is because rustworkx packaging shim will conflict with imports from rustworkx the installed version of rustworkx (which contains the compiled extension).
If you want to run the tests with a specific version of Python, use the test_with_version
target. For example, to launch a test with version 3.11 the command is:
nox --python 3.11 -e test_with_versionIf you just want to run a subset of tests you can pass a selection regex to the
test runner. For example, if you want to run all tests that have "dag" in the
test id you can run: nox -e test -- dag. You can pass arguments directly to the
test runner after the bare --. To see all the options on test selection you
can refer to the stestr manual:
https://stestr.readthedocs.io/en/stable/MANUAL.html#test-selection
If you want to run a single test module, test class, or individual test method
you can do this faster with the -n/--no-discover option. For example:
to run a module:
nox -e test -- -n test_max_weight_matching
or to run the same module by path:
nox -e test -- -n graph/test_nodes.py
to run a class:
nox -e test -- -n graph.test_nodes.TestNodes
to run a method:
nox -e test -- -n graph.test_nodes.TestNodes.test_no_nodes
It's important to note that Nox will be running from the tests/ directory in
the repo, so any paths you pass to the test runner via path need to be relative
to that directory.
When running the visualization tests, each test will generate a visualization
and only fail if an exception is raised by the call. Each test saves the output
image to the current working directory (which if running tests with nox is
tests/) to ensure the generated image is usable. However to not clutter the
system each test cleans up this generated image and by default a test run does
not include any way to view the images from the visualization tests.
If you want to inspect the output from the visualization tests (which is common
if you're working on visualizations) you can set the
RUSTWORKX_TEST_PRESERVE_IMAGES environment variable to any value and this will
skip the cleanup. This will enable you to look at the output image and ensure the
visualization is correct. For example, running:
RUSTWORKX_TEST_PRESERVE_IMAGES=1 nox -e test
will run the visualization tests and preserve the generated image files after the run finishes so you can inspect the output.
As rustworkx-core is a standalone rust crate with it's own public interface it needs it's own testing. These tests can be a combination of doc tests (embedded code examples in the docstrings in the rust code) or standalone tests. You can refer to the rust book on how to add tests:
https://doc.rust-lang.org/book/ch11-01-writing-tests.html
The rustworkx-core tests can be run with:
cargo test
from the rustworkx-core directory.
Rust is the primary language of rustworkx and all the functional code in the libraries is written in Rust. The Rust code in rustworkx uses rustfmt to enforce consistent style. CI jobs are configured to ensure to check this. Luckily adapting your code is as simple as running:
cargo fmtlocally. This will automatically restyle the rust code in rustworkx to match what CI is checking.
An additional step is to run clippy on your changes. You can run it by running:
cargo clippyPython is used primarily for tests and some small pieces of packaging and namespace configuration code in the actual library. black and flake8 are used to enforce consistent style in the python code in the repository. You can run them via Nox using:
nox -e lintThis will also run cargo fmt in check mode to ensure that you ran cargo fmt
and will fail if the Rust code doesn't conform to the style rules.
If black returns a code formatting error you can run nox -e black to automatically
update the code formatting to conform to the style.
Just like with tests building documentation is done via Nox. This will handle compiling rustworkx, installing the python dependencies, and then building the documentation in an isolated venv. You can run just the docs build with:
nox -e docs
which will output the html rendered documentation in docs/build/html which
you can view locally in a web browser.
To build the rustworkx-core documentation you will use rust-doc. You can do this by running:
cargo doc
from the rustworkx-core directory (which is the root of the rustworkx-core
crate. After it's built the compiled documentation will be located in
target/doc/rustworkx_core (which is off the repo root not the rustworkx-core
dir)
You can build and open the documentation directly in your configured default web browser by running:
cargo doc --open
If you have added new methods, functions, or classes, and/or changed any
signatures, type annotations for Python are required to be included in a pull
request. Type annotations are added using type
stub files which
provide type annotations to python tooling which use type annotations. The stub
files are in the rustworkx/ directory and have a .pyi file extension. They
contain annotated signatures for Python functions, stripped of their
implementation. You can find more details on typing in Python at:
- https://mypy.readthedocs.io/en/stable/
- https://typing.readthedocs.io/en/latest/
- https://docs.python.org/3/library/typing.html
Having type annotations is very helpful for Python end-users. Adding annotations lets users type check their code with mypy, which can be helpful for finding bugs when using rustworkx.
Just like with tests for the code, annotations are also tested via Nox.
nox -e stubs
One important thing to note is that if you're adding a new function to the Rust
module you will need to ensure that the signature with annotations is added to
rustworkx/rustworkx.pyi. Then it is also necessary to re-export the annotation
by adding an import line to rustworkx/__init__.pyi in the form:
from .rustworkx import foo as foowhich ensures that mypy is able to find the type annotations when users import
from the root rustworkx package (which is the most common access pattern).
It is important to document any end user facing changes when we release a new version of rustworkx. The expectation is that if your code contribution has user facing changes that you will write the release documentation for these changes. This documentation must explain what was changed, why it was changed, and how users can either use or adapt to the change. The idea behind release documentation is that when a naive user with limited internal knowledge of the project is upgrading from the previous release to the new one, they should be able to read the release notes, understand if they need to update their program which uses rustworkx, and how they would go about doing that. It ideally should explain why they need to make this change too, to provide the necessary context.
To make sure we don't forget a release note or if the details of user facing changes over a release cycle we require that all user facing changes include documentation at the same time as the code. To accomplish this we use the reno tool which enables a git based workflow for writing and compiling release notes.
Making a new release note is quite straightforward. Ensure that you have reno installed with::
pip install -U reno
Once you have reno installed you can make a new release note by running in your local repository checkout's root::
reno new short-description-string
where short-description-string is a brief string (with no spaces) that describes what's in the release note. This will become the prefix for the release note file. Once that is run it will create a new yaml file in releasenotes/notes. Then open that yaml file in a text editor and write the release note. The basic structure of a release note is restructured text in yaml lists under category keys. You add individual items under each category and they will be grouped automatically by release when the release notes are compiled. A single file can have as many entries in it as needed, but to avoid potential conflicts you'll want to create a new file for each pull request that has user facing changes. When you open the newly created file it will be a full template of the different categories with a description of a category as a single entry in each category. You'll want to delete all the sections you aren't using and update the contents for those you are. For example, the end result should look something like::
features:
- |
Added a new function, :func:`~rustworkx.foo` that adds support for doing
something to :class:`~rustworkx.PyDiGraph` objects.
- |
The :class:`~rustworkx.PyDiGraph` class has a new method
:meth:`~rustworkx.PyDiGraph.foo``. This is the equivalent of calling the
:func:`~rustworkx.foo` function to do something to your
:class:`~rustworkx.PyDiGraph` object, but provides the convenience of running
it natively on an object. For example::
from rustworkx import PyDiGraph
g = PyDiGraph.
g.foo()
deprecations:
- |
The ``rustworkx.bar`` function has been deprecated and will be removed in a
future release. It has been superseded by the
:meth:`~rustworkx.PyDiGraph.foo` method and :func:`~rustworkx.foo` function
which provides similar functionality but with more accurate results and
better performance. You should update your calls
``rustworkx.bar()`` calls to use ``rustworkx.foo()`` instead.You can also look at other release notes for other examples.
You can use any sphinx feature in them (code sections, tables, enumerated lists, bulleted list, etc) to express what is being changed as needed. In general you want the release notes to include as much detail as needed so that users will understand what has changed, why it changed, and how they'll have to update their code.
After you've finished writing your release notes you'll want to add the note
file to your commit with git add and commit them to your PR branch to make
sure they're included with the code in your PR.
If you need to link to an issue or other Github artifact as part of the release note this should be done using an inline link with the text being the issue number. For example you would write a release note with a link to issue 12345 as:
fixes:
- |
Fixes a race condition in the function ``foo()``. Refer to
`#12345 <https://github.com/Qiskit/rustworkx/issues/12345>`__ for more
details.After release notes have been added if you want to see what the full output of
the release notes. Reno is used to combine the release note yaml files into a
single rst (ReStructuredText) document that
sphinx will then compile for us as part
of the documentation builds. If you want to generate the rst file you
use the reno report command. If you want to generate the full rustworkx
release notes for all releases (since we started using reno during 0.8) you just
run::
reno report
but you can also use the --version argument to view a single release (after
it has been tagged::
reno report --version 0.8.0
Building the release notes is part of the standard rustworkx documentation
builds. To check what the rendered html output of the release notes will look
like for the current state of the repo you can run: nox -e docs which will
build all the documentation into docs/_build/html and the release notes in
particular will be located at docs/_build/html/release_notes.html
After you've submitted a pull request to rustworkx it will need to pass CI and be reviewed by an approved by a core team reviewer. CI runs get triggered automatically when your pull request is opened and on every subsequent commit made to your pull request's branch. Code review however may take some time, sometimes even weeks or months, there are many new pull requests opened every day and limited number of reviewers available, and while every proposed change is a valuable addition to the project not everything is the highest priority. You can help this process move more quickly by actively reviewing other open PRs. While only members of the rustworkx core team have permission to provide final approval and mark a PR as ready for merging, reviewing code is open to everyone and all reviews are welcome and extremely valued contributions. Helping with code review also helps reduce the burden on the core team and enables them to review code faster.
The code review process is a bit of back and forth where you will receive feedback and questions about your proposed changes to the project. You will likely have multiple rounds of feedback with suggestions or changes requested before approval. Please do not get discouraged as this is normal and part of ensuring the quality of the rustworkx project and even what first appears as a straightforward or simple change might have larger implications that aren't obvious at first. If you receive feedback feel free to request re-review from reviewers after you've adjusted your PR based on the comments received.
Another thing to keep in mind is that CI time is a constrained resource and not
infinite. While waiting for review and approval it is not necessary to keep your
PR branch up to date on every change to the main branch. Doing it periodically
is fine to make sure there are no regressions as the codebase changes, but
doing it too often will just needlessly waste CI resources. This will contribute
to resource starvation on CI, slowing down total throughput for the project. If
possible try to bundle updating your branch to the current HEAD on the main
branch with other changes made to the PR branch (like making adjustments from
code review). This will result in a single CI run instead of doing standalone
updates with no code changes.
Once your PR has the necessary approvals it will be tagged with the automerge
tag. This is a signal to the mergify bot that the PR has
been approved and is ready for merging. The mergify bot will then enqueue the
PR onto its merge queue. At this point the process of updating a PR to the
current HEAD of the main branch is fully automated and once CI passes mergify
will merge the PR automatically. In an effort to conserve CI resources and
maximize throughput the mergify bot will only update a PR when it's next in the
merge queue. It might appear as activity on your PR is idle at this point, but
this likely just means the mergify merge queue is deep and/or CI has a backlog.
Do not manually update a PR branch to HEAD on the main branch after it
has the necessary approvals and is tagged as automerge unless it has a merge
conflict or has a failed CI run. Doing so will just waste CI resources and
delay everything from merging, including your PR.
The stable branch is intended to be a safe source of fixes for high-impact bugs, documentation fixes, and security issues that have been fixed on main since a release. When reviewing a stable branch PR, we must balance the risk of any given patch with the value that it will provide to users of the stable branch. Only a limited class of changes are appropriate for inclusion on the stable branch. A large, risky patch for a major issue might make sense, as might a trivial fix for a fairly obscure error-handling case. A number of factors must be weighed when considering a change:
- The risk of regression: even the tiniest changes carry some risk of breaking something, and we really want to avoid regressions on the stable branch.
- The user visibility benefit: are we fixing something that users might actually notice, and if so, how important is it?
- How self-contained the fix is: if it fixes a significant issue but also refactors a lot of code, it’s probably worth thinking about what a less risky fix might look like.
- Whether the fix is already on main: a change must be a backport of a change already merged onto main, unless the change simply does not make sense on main.
Normally only bug fixes or non-code changes are allowed on a stable branch, the primary exception to this is adding support for new python versions. If a new python version is released backporting that feature change with that new support is an acceptable backport.
In rustworkx at least until the 1.0 release we only maintain a single stable branch at a time for the most recent minor version release.
In the normal case to backport a pull request all that needs to be done is
to tag it as stable-backport-potential, this will signal the
mergify bot that the PR should be backported after it
merged. Once a PR tagged as stable-backport-potential merges mergify will
automatically open a new PR backporting it to the stable branch.
If the mergify approach doesn't work for some reason and you need to manual backport a PR this can be done with the following procedure. When backporting a patch from main to stable, we want to keep a reference to the change on main. When you create the branch for the stable PR, use:
$ git cherry-pick -x $main_commit_id
However, this only works for small self-contained patches from main. If you need to backport a subset of a larger commit (from a squashed PR, for example) from main, do this manually. In these cases, add:
Backported from: #main pr number
so that we can track the source of the change subset, even if a strict cherry-pick doesn't make sense.
If the patch you’re proposing will not cherry-pick cleanly, you can help by resolving the conflicts yourself and proposing the resulting patch. Please keep Conflicts lines in the commit message to help review of the stable patch.