This monorepo contains the design system of MagicBell.
Warning:
The packages in this repo aren't ready. We're actively working on them, and breaking changes are still to be expected while we iron out the look & feel of their visual appearance, as well as code signatures.
Expect breaking changes on every release until we reach
v1.0.0
.
The development environment for this repo supports node:v18
and up. We use yarn
as package manager.
Run the following commands in your terminal to get a copy of this repo, and install required dependencies.
git clone git@github.com:magicbell/darkmagic.git
cd darkmagic
yarn
When working on the packages, you'll want to run the builder in watch mode. You can do so by:
yarn start
This builds the packages, and recompiles all affected ones when you make changes in the packages
folder.
We have a single storybook instance available for stories in all packages. Run it with:
yarn start:storybook
After that, visit http://localhost:6006. Storybook will hot-reload on any change that you make in /packages
. Make sure that you're running yarn start
in another terminal, when you have the feeling that you're dealing with stale data.
We have a playroom instance available for quick prototyping. Run it with:
yarn start:playroom
After that, visit http://localhost:9000. Playroom will hot-reload on any change that you make in /packages
. Make sure that you're running yarn start
in another terminal, when you have the feeling that you're dealing with stale data.
There is a vite based example application located in the /example folder. This gives one the opportunity to try out use cases in a production like environment. Run it with:
yarn start:example
After that, visit http://localhost:3000. The example app will hot-reload on any change that you make in /packages
. Make sure that you're running yarn start
in another terminal, when you have the feeling that you're dealing with stale data.
Code quality is set up for you with eslint
, prettier
, husky
, and lint-staged
. Please keep the pre-commit hooks enabled.
You can take advantage of invariant, warning and __DEV__
to add development-only warnings which won't end up in our production bundles.
import invariant from 'tiny-invariant';
import warning from 'tiny-warning';
invariant(truthyValue, 'This should not throw!');
invariant(falsyValue, 'This will throw!');
warning(truthyValue, 'This should not log a warning');
warning(falsyValue, 'This should log a warning');
if (__DEV__) {
// this is excluded from the production bundle
}
Tests are setup using vitest. They can be run once with coverage with yarn test
or use yarn test:watch
to run them in watch mode.
yarn test:watch
When contributing changes, make sure to document them in a changeset. You'll usually do this when you're ready to push the changes and create a pull-request. To do so, run:
yarn changeset
Changeset
will ask you for a change description, and how different packages are affected. Your answers help us determine if packages should be bumped in version, and if it should be a patch
, minor
, or major
version bump. The description you provide may end up in our changelogs.
Please review, polish, and commit the files after completing the steps.
Publishing new package versions is an automated process managed via the release
workflow. To trigger a release, push a commit containing changesets (yarn changeset
) to main
. This will trigger the bot to open or update a pull-request named next release
. Once that pull request gets merged, the bot will publish the new versions to npm and create the release notes on GitHub.
Manual publishing is done in two steps:
yarn changeset:version
This command consumes the changesets as collected in .changeset, and proposes changelogs and version bumps. Please review the proposed changes, and polish the changelogs. Make sure that breaking changes result in major version bumps.
Commit the change with git add . && git commit -m 'version packages'
, and move on to the next step to publish.
yarn changeset:release
This will publish all changed packages to npm, and tag the last commit. Please don't commit anything between the release commit, and the publish action. These command are separate to enable you to check if the release commit is accurate.