Welcome! This readme assumes you have already listened to the 311-data pitch, and gone through the basic onboarding. The following will be more geared towards the programming side of 311-data and understanding the rules of the road before getting your development environment setup. If you run into any problems, please submit a new issue.
But before you begin, please turn on two-factor authentication for GitHub. We require 2FA to be in place for all contributors to the project.
First, here is a high-level look at our architecture:
The stuff in blue is external to the stack (services owned by 3rd parties). The stuff in the AWS box are all server containers running Python (i.e., the 311 Data Server). The green is the React SPA which users interact with (i.e., the 311 Data Client). We have set this up as a monolithic project in GitHub and whether you develop on the server or the client you will follow the same process.
For JavaScript developers looking to contribute to the 311 Data client, start by reading how to set up a local client environment
For Python developers looking to contribute to the 311 Data server, start by reading how to set up a local API server
For context on how we do branches, start by reviewing the 311 Data release process
For development we use feature branching to ensure easy collaboration. There aren't any rules to branch naming or how many branches you are allowed to have, but the recommended convention would look like issueId-Prefix-MinimalDescription
For example, a documentation branch could look like 138-DOC-OnboardingUpdate.
Read more about feature branching here.
Building on feature branching, we treat the dev branch as the main contribution branch. Pull requests to this branch should be as frequent as developers are closing issues (Hopefully very frequent!). Pushes to master will be much less frequent and will be handled by administrators. With this workflow, master will have an extra layer of protection and should always represent a working version of the application.
In other words, whenever you are about to start on a new feature, checkout your branch based off of the dev branch. Your command would look something like git checkout -b 567-BACK-NewEndpoint dev. See this stackoverflow post for more context.
We use Github Actions to run our continuous integration (CI). These actions include status checks that run whenever you submit a pull request to dev or master. When you submit a PR, Github will run a set of operations to build and test all or part of the codebase. If any of these steps fail, the pull request will not be allowed to be merged until they are fixed. From the pull request UI you can find the reason an operation may have failed in the status checks section towards the bottom.
If you want to look at our setup, check out the "Actions" tab in Github, as well as the workflows directory, which contains the code that Github runs when actions are triggered.
In addition to status checks, PR's are required to have at least one reviewer before being merged into dev or master.
CI Is driven by tests, they help instill confidence in pull requests because a developer can say "All the status checks pass and my new tests pass so the PR is safe to merge" When contributing new features, it is most ideal to write at least 4 tests targeting your code.
- One for the "happy path": Test the endpoint/feature in the way it is intended to be used
- One for the "extreme path": Test with extreme inputs/characteristics (What if I use 10,000 XYZ)
- One for the "negative path": Test with strange input, (What if I send characters to a function that expects integers)
- One for the "null path": Test with empty params/nothing/emptiness
Our front end tests are run through Cypress and our backend tests are run through Pytest.