How to Contribute
We'd love to accept your patches and contributions to this project. We do have some guidelines to follow, covered in this document, but don't worry about (or expect to) get everything right the first time! Create a pull request and we'll nudge you in the right direction. Please also note that we have a code of conduct to make Cirq an open and welcoming environment.
Contributor License Agreement
Contributions to this project must be accompanied by a Contributor License Agreement. You (or your employer) retain the copyright to your contribution; this simply gives us permission to use and redistribute your contributions as part of the project. Head over to https://cla.developers.google.com/ to see your current agreements on file or to sign a new one.
You generally only need to submit a CLA once, so if you've already submitted one (even if it was for a different project), you probably don't need to do it again.
Pull Request Process and Code Review
All submissions, including submissions by project members, require review. We use GitHub pull requests for this purpose. GitHub Help has information on using pull requests.
The preferred manner for submitting pull requests is for users to fork the Cirq repo and then use a branch from this fork to create a pull request to the main Cirq repo.
The basic process for setting up a fork is
- Fork the Cirq repo (Fork button in upper right corner of
Forking creates a new github repo at the location
USERNAMEis your github id. Use the directions on the development page to download a copy to your local machine. You need only do this once.
- Checkout master and create a new branch from this master
git checkout master -b new_branch_name
new_branch_nameis the name of your new branch.
- Do your work and commit your changes to this branch.
- If you have drifted out of sync with the master from the
main cirq repo you may need to merge in changes. To do this,
first update your local master and then merge the local master
into your branch:
You may need to fix merge conflicts for both of these merge commands.
# Track the upstream repo (if your local repo hasn't): git remote add upstream https://github.com/quantumlib/Cirq.git # Update your local master. git fetch upstream git checkout master git merge upstream/master # Merge local master into your branch. git checkout new_branch_name git merge master
- Finally, push your change to your clone
git push origin new_branch_name
- Now when you navigate to the cirq page on github, https://github.com/quantumlib/cirq you should see the option to create a new pull request from your clone repository. Alternatively you can create the pull request by navigating to the "Pull requests" tab in the page, and selecting the appropriate branches.
- The reviewer will comment on your code and may ask for changes, you can perform these locally, and then push the new commit following the same process as above.
Development Environment Setup
Please refer to our development page for instructions on setting up your local development environment.
Code Testing Standards
When a pull request is created or updated, various automatic checks will run to ensure that the change won't break Cirq and meets our coding standards.
Cirq contains a continuous integration tool to verify testing. See our development page on how to run the continuous integration checks locally.
Please be aware of the following code standards that will be applied to any new changes.
- Tests. Existing tests must continue to pass (or be updated) when new changes are introduced. We use pytest to run our tests.
Code should be covered by tests.
We use pytest-cov to compute
coverage, and custom tooling to filter down the output to only include new or
changed code. We don't require 100% coverage, but any uncovered code must
be annotated with
# coverage: ignore. To ignore coverage of a single line, place
# coverage: ignoreat the end of the line. To ignore coverage for an entire block, start the block with a
# coverage: ignorecomment on its own line.
Code should meet common style standards for python and be free of error-prone
constructs. We use pylint to check for lint.
To see which lint checks we enforce, see the
dev_tools/conf/.pylintrc file. When pylint produces
a false positive, it can be squashed with annotations like
# pylint: disable=unused-import.
Code should have type annotations.
We use mypy to check that type annotations are correct.
When type checking produces a false positive, it can be ignored with
# type: ignore.
Request For Comment Process for New Major Features
For larger contributions that will benefit from design reviews, please use the Request for Comment process.
Please refer to our notebooks guide on how to develop iPython notebooks for documentation.