CONTRIBUTING.md

May 9, 2022 · View on GitHub

Contributing

We welcome all kinds of contributions, there are several ways to contribute depending on what you're interested in:

Framework: If you are interested in the framework process of federated learning and have a certain understanding of the basic concepts of federated learning, you are very welcome to supplement or complete the new federated learning features. In addition, you can also optimize and improve existing processes and codes. We strongly recommend that you open an issue for discussion before developing new features, so that you might get a more suitable solution.
Algorithm: If you are interested in algorithms such as privacy or secure aggregation used in federated learning, then we welcome you to develop new algorithms to integrate into the framework. We have a complete set of extension mechanisms to develop new algorithms without needing to know the framework.
Docs: If you are interested in helping us improve the developer experience, we also welcome you to supplement some documents, such as detailed instructions to the API, or corrections of errors in the documents.
Tutorials: If you are very interested in the application of federated learning, you can try to provide examples of application scenarios, new federated learning datasets, or federated models, etc., which can help other users better understand federated learning, and its application scenarios.

Of course not limited to the above, please read the guidelines below carefully before submitting.

All submissions, including submissions by project members, require review. We use GitHub pull requests for this purpose. Consult GitHub Help for more information on using pull requests. One of the responsible code owners will then review your contribution later.

Notice: You must comply with the DCO agreement when submitting, refer to Legal Notice bellow.

Bugs or Issues

You are very welcome to point out any errors in the code or documentation, you can directly submit a PR to fix it. Also, you can illustrate bugs and features requests via the GitHub issues. If you need help, ask your questions via the mailing list bellow.

mail:

Code Style

In order to better maintain the consistency of the code and make it easier for developers to understand, please following the code guidelines bellow:

Python style: PEP 8 – Style Guide for Python Code
Protocol buffers style： Protocol Buffers Style Guide
When submitting new features or fixing bugs, please attach the corresponding unittest to ensure the correctness of the code. Follow the python built-in unittest.TestCase

PR Checklist

When you are ready to submit a PR, please check this list before submitting PR：

Check to make sure that every commit has signed the DCO agreement, otherwise check the method here to fix it.
Ensure that the code is statically checked by pylint, flake8 tool, you can install the tools by:
```
pip install pylint flake8
```
Ensure that the unittest is tested by the pytest tool, you can install the tool by:
```
pip install pytest
```
Use bazel build to make sure the project compiles successfully.

For simplicity, you can also execute the tox command in the project root directory to complete the checks 2~4 items. The process is as follows:

pip install tox

cd federated-learning/

tox

After the above checklist is passed, then you can submit the PR.

Legal Notice

We have adopted the DCO(Developer Certificate of Origin) as the contributor agreement, so be sure to understand the usage of the agreement before submitting. The usage method is as follows:

In particular, you must sign-off on every commit, otherwise the entire PR will not be accepted, the format as follows(with a single line in the commit message):

Signed-off-by: Full Name <email>