Skip to content

Latest commit

 

History

History
136 lines (80 loc) · 6.55 KB

CONTRIBUTING.md

File metadata and controls

136 lines (80 loc) · 6.55 KB
Raw

Contributing

Thanks for being willing to contribute! 🙏

Working on your first Pull Request (PR)? You can learn how from this free series How to Contribute to an Open Source Project on GitHub

Open issues

Please check out the the open issues. Issues labelled Help Wanted and Good First Issue are especially good to help with.

Contributing doesn’t have to be in code! Simply answering questions in open issues, or providing workarounds, is just as important a contribution as making pull requests.

Opening a Pull Request

Pull requests are welcome for this repo! Bugfixes will always be accepted, though in some cases some small changes may be requested.

However, if adding a feature or breaking change, please open an issue first to discuss. This ensures no time or work is wasted writing code that won’t be accepted to the project (see Project Goals). Undiscussed feature work may be rejected at the discretion of the maintainers.

Setup

  1. Install pnpm
  2. Fork this repo and clone your copy locally
  3. Run pnpm i to install dependencies

Writing code

Create a new branch for your PR with git checkout -b your-branch-name. Add the relevant code as well as docs and tests. When you push everything up (git push), navigate back to your repo GitHub and you should see a prompt to open a new PR.

While best practices for commit messages are encouraged (e.g. start with an imperative verb, keep it short, use the body if needed), this repo doesn’t follow any specific guidelines like Conventional Commits. Clarity is favored over strict rules. Changelogs are generated separately from git (see the Changelogs section

When working locally, run:

npm run dev

This will compile the code as you change automatically.

Writing the PR

Please fill out the template! It’s a very lightweight template 🙂.

Use Test-driven Development!

Contributing to this library is hard-bordering-on-impossible without a test-driven development (TDD) strategy. If you’re new to this, the basic workflow is:

  1. First, write a test that fully outlines what you’d like the output to be.
  2. Next, make sure this test fails when you run npm test (yes, fails!)
  3. Then, make changes to src/ until the tests pass.

Reasoning about code generation can be quite difficult until you “invert your thinking” and approach it output-first. Adopting TDD can turn very unclear/abstract problems into concrete ones with clear steps to resolution.

✨ When starting any task, write a failing test first! ✨

Updating snapshot tests

To add a schema as a snapshot test, modify the /scripts/download-schemas.ts script with a path to download. There are both single-file schemas as well as multi-file schemas.

Generating types

It may be surprising to hear, but generating TypeScript types from OpenAPI is opinionated! Even though TypeScript and OpenAPI are very close relatives, both being JavaScript/JSON-based, they are nonetheless 2 different languages and thus there is always some room for interpretation. Likewise, some parts of the OpenAPI specification can be ambiguous on how they’re used, and what the expected type outcomes may be (though this is generally for more advanced usecasees, such as specific implementations of anyOf as well as discriminator and complex polymorphism).

All that said, this library should strive to generate the most predictable TypeScript output for a given schema. And to achieve that, it always helps to open an issue or discussion to gather feedback.

Opening a PR

When opening a pull request, make sure all of the following is done:

  • Tests are added
  • Build passes (npm run build)
  • Tests pass (npm test)
  • Linting passes (npm run lint)

Lastly, be sure to fill out the complete PR template!

Changelogs

The changelog is generated via changesets, and is separate from Git commit messages and pull request titles. To write a human-readable changelog for your changes, run:

npx changeset

This will ask if it’s a patch, minor, or major change (semver), along with a plain description of what you did. Commit this new file along with the rest of your PR, and during the next release this will go into the official changelog!

Testing

This library uses Vitest for testing. There’s a great VS Code extension you can optionally use if you’d like in-editor debugging tools.

Running tests

💡 The tests test the production build in dist/. Be sure to run npm run build before running tests (or keep npm run dev running in the background, which compiles as-you-work)!

To run the entire test suite once, run:

npm test

To run an individual test:

npm test -- [partial filename]

To start the entire test suite in watch mode:

npx vitest

Running linting

To run ESLint on the project:

npm run lint

Updating snapshot examples

⚠️ This may break tests if schemas have been updated

npm run update:examples

Unit tests or snapshot tests?

This library has both unit tests (tests that test a tiny part of a schema) and snapshot tests (tests that run over an entire, complete schema). When opening a PR, the former are more valuable than the latter, and are always required. However, updating snapshot tests can help with the following:

  • Fixing bugs that deal with multiple schemas with remote $refs
  • Fixing Node.js or OS-related bugs
  • Adding a CLI option that changes the entire output

For most PRs, snapshot tests can be avoided. But for scenarios similar to the ones mentioned, they can ensure everything is working as expected.