We use Yarn for this repository (v1.21 and above), with Node.js v10 and above.
For Node.js we recommend using nvm
There is one yarn lock in this repo at the root level, this is like package-lock.json for npm users.
This file is committed it ensures we use the same exact version of the packages in CI.
DO NOT USE yarn to install dependencies, as this will modify the yarn.lock.
=> Use yarn install:all, which includes the --frozen-lockfile flag to prevent unnecessary changes to yarn.lock.
We use Workspaces as explained in our GETTING STARTED guide.
The best way is usually to manually modify the package.json in the correct workspace (root, linode-manager or @linode/api-v4),
then run yarn. This should update the lockfile with only the required changes.
If the resulting changes to yarn.lock are extensive (which can be the case if Yarn detects that it can resolve warnings in
subdependencies), consider pushing a separate PR with these changes so that individual PRs are kept small and manageable.
If you see that 1 new dependency you added included a lot of subdependencies, with older versions of the dependencies we already have, check that the new package is not outdated
Prefer yarn run over npx
yarn run <command> will execute a command in the root workspace.
yarn run workspace <workspace> run <command> to execute it in a specific project.
Built-in commands do not require run.
In the case of yarn add running it in the root directory will issue a warning, this is normal and just to make sure that you want to install the dependency there and not in one of the workspaces. (you need to add -W to ignore it)
yarn install:all: installs dependencies, without touching the lock.
yarn postinstall: is called automatically after any yarn install, applies patches to the packages installed.
yarn clean: removes all dependencies installed (equivalent to rm -rf node_modules).
yarn build builds the SDK and the production build of the Cloud Manager. Add the --analyze flag to produce a report of the bundle size and content in packages/manager/bundle_analyzer_report.html.
yarn build:sdk: builds the SDK.
yarn start:all: starts the development server on localhost:3000 with hot reload enabled (both SDK and Manager).
yarn up installs everything, builds the sdk, runs yarn start:all
yarn test: runs the unit tests (with a fresh clone of the repo you must run yarn build:sdk before running the tests)
yarn cy:e2e: runs the e2e tests (you must have the app served on localhost:3000 with yarn up)
yarn storybook:e2e: runs the storybook tests (you must have the storybook server started with yarn storybook)
More on testing commands in testing docs