nav_order |
---|
99 |
Hi there, thanks for checking out our repo!
skuba is a toolkit for developing TypeScript backend applications and packages at SEEK. While third-party contributions are certainly welcome, this project is primarily driven by our internal priorities and technical guidelines.
SEEKers: this repo is public, so don't commit or post anything that isn't ready for the entire world to see.
skuba is documented through its README, along with some targeted topics under the docs directory. We maintain a changelog and release notes on GitHub, and distribute it as an npm package.
Submit an issue if you have a question, feature request or bug report.
If you work at SEEK, start a discussion in #skuba-support.
Feel free to create a pull request for trivial fixes and improvements.
For more substantial features, please submit an issue first. This lets us evaluate whether the feature fits the direction of the project and discuss possible approaches.
If you work at SEEK, start a discussion in #sig-backend-tooling.
skuba is predominantly tested on macOS and Linux. If you're on Windows, we recommend the Windows Subsystem for Linux.
First, some JavaScript tooling:
- Node.js LTS
- pnpm
Next, install npm dependencies:
pnpm install
We use GitHub flow.
Create a new branch off of the latest commit on the default branch:
git fetch origin
git switch --create your-branch-name origin/main
Develop, test and commit your changes on this branch. (Make sure to include the appropriate changeset.)
git add --all
git commit
Finally, push your branch to GitHub and create a pull request:
git push --set-upstream origin your-branch-name
If you don't have push access, you may need to fork the repo and push there instead:
git remote add fork git@github.com:your-username/skuba.git
git push --set-upstream fork your-branch-name
A maintainer will get to your pull request and review the changes. If all is well, they will merge your pull request into the default branch.
You may find it easier to develop alongside unit tests:
pnpm test --watch
Format your code once you're happy with it:
pnpm format
We run linting and testing in CI, but consider running these commands locally for a faster feedback loop:
pnpm lint
pnpm test
Our validate GitHub Actions workflow also initialises each built-in skuba template and runs through a set of CLI commands. This can be reproduced locally, but keep in mind that the script is fairly slow and you'll have to manually clean up afterwards.
# greeter | koa-rest-api | ...
pnpm test:template greeter
# clean up temporary sibling directory
rm -fr ../tmp-greeter
Snapshots in templates
can be updated with the following command but note that manual clean up is necessary afterwards
# greeter being name of template
pnpm test:template:updateSnapshot greeter
# clean up temporary sibling directory
rm -fr ../tmp-greeter
If you want to try out the skuba CLI on itself,
a pnpm skuba
script is configured:
# Prints available commands.
pnpm skuba
# Prints version from local package.json.
pnpm skuba version
# Builds skuba using itself.
pnpm skuba build
If you want to try out the skuba CLI on another local repo, use pnpm link:
# Do this once upfront.
pnpm link --global
# `pnpm link` points to the JavaScript output in `./lib`.
# This means you'll need to rebuild skuba on every code change 😔.
pnpm build
# Run skuba commands against the other repo.
skuba version
# Avoid command confusion after you're done.
pnpm uninstall --global skuba
We use Changesets to manage package releases. You'll see a 🦋 bot gliding around pull requests.
You should write a changeset if you are changing the public skuba interface, which includes:
- API for Node.js build and test code
- CLI commands
- Config presets
- Template code and documentation
- npm dependencies
On the other hand, a changeset is not necessary for:
- Documentation like the README
- Internal refactoring that preserves the existing interface
- npm dev dependencies
pnpm changeset
The Changesets CLI is interactive and follows semantic versioning:
- Patch release
0.0.X
: fixes or tweaks to existing functionality - Minor release
0.X.0
: new, backwards-compatible functionality - Major release
X.0.0
: backwards-incompatible modification
We humour several transgressions to this versioning scheme in practice:
-
Breaking changes to
skuba lint
should be downgraded to minor.It's not feasible for us to semantically version based on whether
skuba lint
will pass or fail, especially given that lint rules can change between minor and patch versions of transitive ESLint dependencies. The general thought here is that changes that can affect the runtime behaviour of your project should be major, while changes to build-time validation of a project should not be major.We also support autofixes to ease adoption of lint rule changes.
-
Breaking changes in TypeScript upgrades should be downgraded to minor.
TypeScript does not follow semantic versioning, and its point releases generally come with breaking changes. These are typically edge cases that would not affect a typical SEEK project.
In the event that a new compiler rule presents significant challenges to existing SEEK projects, we may turn off the rule by default to revert its impact, or publish a major with detailed guidance on how to comply with or disable the rule.
-
Changes to built-in templates should be downgraded to patch.
Release notes and package versioning are most pertinent to existing projects. Once you've run
skuba init
, updates to built-in templates are largely inconsequential to your project, and mostly serve as a footnote to communicate best current practices.
Prefix your changeset title with a scope:
.
This makes it easy to eyeball which part of skuba a change relates to.
test: Fix command
template: Add next steps to READMEs
template/koa-rest-api: Switch to Express
format, lint: Introduce new ESLint rule
The Changesets CLI will generate a Markdown file under .changeset, which you should include in your pull request. It doesn't need to be part of the same commit as the rest of your changes. Feel free to manually edit this file to include more details about your change.
When a pull request with a changeset is merged,
our CI workflow will create a new Version Packages
PR.
The changesets are used to infer the next semantic version and to update the changelog.
This PR may be left open to collate multiple changes into the next version. A maintainer will merge it once ready, and our release GitHub Actions workflow will publish the associated GitHub release and npm package version.
Prereleases can be created on demand via seek-oss/changesets-snapshot.
Manually run the Snapshot workflow for the main
branch in GitHub Actions to publish a new snapshot version to npm.