Skip to content

Latest commit

 

History

History
205 lines (131 loc) · 10.5 KB

CONTRIBUTING.md

File metadata and controls

205 lines (131 loc) · 10.5 KB
Raw

Contributing to Lerna

Please note that this project is released with a Contributor Code of Conduct. By participating in this project you agree to abide by its terms.

Installing the correct versions of Node, NPM and necessary dependencies

We strongly recommend using Volta to manage your node and npm installation seamlessly and automatically. If you are already using volta, then you will already be using the correct versions of node and npm when working on the lerna repo.

In addition to being used by our core contributors, volta is also used by our Github Actions workflows, so that everything stays automatically in sync when it comes to the primary node and npm versions.

If you do not wish to use volta for whatever reason, that is absolutely fine, but you will be responsible for ensuring that you align your machine's global installation of node and npm with what we have specified in the volta config within the root package.json file in the repo (and keep it aligned as things change over time).

You can always run node --version and npm --version from within the repo to check that your currently installed versions match those specified in the volta config in the root package.json.

Once you have done that, to get started with the repo simply run:

git clone [email protected]:lerna/lerna.git && cd lerna
npm ci # given this is a clean install on an existing project, npm ci can be used

Code Structure

Currently, the source is split up into a few categories:

  • e2e: latest test code which runs the lerna CLI as close to how a user does as possible
  • integration: legacy test code which exists somewhere between unit and e2e testing in terms of abstraction. New e2e tests should be preferred over new tests here because they are higher value.
  • libs: source code which gets composed into packages for publishing, or which assists with other things such as unit and e2e testing
  • packages: the packages which get published to npm
  • tools: utility scripts, node_module patches etc which help with maintaining the repository
  • website: source code for https://lerna.js.org, which ultimately gets published to Github pages from https://github.com/lerna/website for historical reasons

Submission Guidelines

Submitting an Issue

Before you submit an issue, please search the issue tracker. An issue for your problem may already exist and has been resolved, or the discussion might inform you of workarounds readily available.

We want to fix all the issues as soon as possible, but before fixing a bug we need to reproduce and confirm it. Having a reproducible scenario gives us wealth of important information without going back and forth with you requiring additional information, such as:

  • the output of npx lerna info
  • yarn.lock or package-lock.json
  • and most importantly - a use-case that fails

A minimal reproduction allows us to quickly confirm a bug (or point out a coding problem) as well as confirm that we are fixing the right problem.

We will be insisting on a minimal reproduction in order to save maintainers' time and ultimately be able to fix more bugs. Interestingly, from our experience, users often find coding problems themselves while preparing a minimal reproduction repository. We understand that sometimes it might be hard to extract essentials bits of code from a larger codebase, but we really need to isolate the problem before we can fix it.

You can file new issues by filling out our issue form.

Submitting a PR

This project follows GitHub's standard forking model. Please fork the project to submit pull requests.

Commands

Run Unit Tests

# Run all unit tests
npm test

# Watch all unit tests for changes
npm test -- --watch

# Test a specific project or package
npx nx test core
npx nx test lerna
# etc

# Watch a specific project or package's tests
npx nx test core --watch
npx nx test lerna --watch
# etc

Run Integration Tests

npm run integration

# test a specific file
npm run integration -- --testFile lerna-add.spec.ts

# watch for changes
npm run integration -- --watch

# watch a specific file
npm run integration -- --testFile lerna-add.spec.ts --watch

Linting

npm run lint

It's also a good idea to hook up your editor to an ESLint extension (such as vscode-eslint).

To fix lint errors from the command line:

npm run lint -- --fix

Local CLI Testing

If you want to test out Lerna on local repos, you can leverage verdaccio as a local npm registry.

You will need two different terminal windows for this as one of them will contain the long-running serve command of the verdaccio instance (your local npm registry) which you will be publishing to.

  • Run npm run local-registry start in Terminal 1 (keep it running)
  • Run npm adduser --registry http://localhost:4873 in Terminal 2 (real credentials are not required, you just need to be logged in. You can use test/test/[email protected].)
  • Run npm --registry=http://localhost:4873/ run lerna-release 999.9.9 --local in Terminal 2 - you can choose any nonexistent version number here, but it's recommended to use something which is different from the current major

You can then install your local version of lerna wherever you want by leveraging the --registry flag on the npm/npx client.

E.g. you could start a new lerna workspace with your new version

cd /some/path/on/your/machine
npx --registry=http://localhost:4873/ [email protected] init
npm --registry=http://localhost:4873/ install
npx lerna --version # 999.9.9

NOTE: After you finish with local testing don't forget to stop the local registry (e.g. closing the Terminal 1) and disabling the local registry using npm run local-registry disable. Keeping the local registry enabled will change your lock file resolutions to localhost:4873 on the next npm install. You can also run npm run local-registry clear to clean all packages in that local registry.

NOTE: To use this newly published local version, you need to make a new workspace or change all of your target packages to this new version, eg: "lerna": "^999.9.9", and re-run npm install in your testing project.

Run E2E Tests

NOTE: You will need to have pnpm installed at version 8.10.2 or later in order to run the full suite of e2e tests

In addition to our lower level testing, we also have a suite of e2e tests which actually publish our packages to a locally running npm registry (using verdaccio in the same way as described in the section above) and use the lerna CLI directly. These are our most valuable tests because they get us as close as possible to the experience of our users.

Because of this high-value nature of the tests, they are also much slower than unit tests. Therefore they are split up into different e2e projects in the workspace which can be run independently and can benefit from more granular caching (thanks to Nx).

The core concepts of building and publishing the packages to a local registry and then invoking the lerna CLI just look a user would remain the same across all areas, but there are some slightly different instructions for the task-runner tests.

All Projects (except e2e-run-task-runner)

To run the e2e tests for a particular project, such as e2e/info, which tests the lerna info CLI command, you can run:

npx nx e2e e2e-info

You can forward arguments onto the underlying jest process like so:

# This will cause jest to only match against tests which have `qqqq` in their `describe()` or `it()` description (thanks to -t), and update the snapshots (thanks to -u)
npx nx e2e e2e-info -t qqqq -u

NOTE: The building, versioning and publishing of the packages will be the same regardless of the jest flags passed

Testing the task-runner (e2e-run-task-runner)

Because the task-runner itself handles spawning multiple nested child processes in node, it becomes hard to wrap that in further node child processes for the purposes of collecting the stdout and stderr outputs in order to assert things about them in test files. This is because node offers no guarantees around the ordering of writes to those stdout and stderr streams, meaning that, even though all the CLI output lines are the same as what a user would see, the ordering of those lines could be non-deterministic in our tests.

Therefore in our task-runner tests we use bash to invoke the lerna CLI and first write the outputs to disk. Then our node tests read those files and assert things about them. We've found by doing it this way we remove most if not all of the non-determinism from the tests.

To run the e2e tests for the task runner, you first need to make sure the local registry is running and has had the packages published to it. For that you can run:

npx nx prepare-for-e2e e2e-run-task-runner

Then you can actually execute the tests by running the bash script:

e2e/run/task-runner/src/run-tests.sh

This bash script takes the name of a subdirectory within e2e/run/task-runner/src as an argument, allowing you to focus on a smaller subset of tests. E.g. to just run the tests in e2e/run/task-runner/src/env-files, you can run:

e2e/run/task-runner/src/run-tests.sh env-files

If you pass --update-snapshots to the shell script it will run jest with the -u option to update any existing snapshots.

E.g.

e2e/run/task-runner/src/run-tests.sh --update-snapshots # to update all test snapshots
e2e/run/task-runner/src/run-tests.sh env-files --update-snapshots # to update just the snapshots env-files

Releasing

If you are a member of Lerna's GitHub org and have read-write privileges in Lerna's npm org with 2-factor auth enabled, congratulations, you can cut a release!

You'll need to set up a local .env file in the repo root to provide the required environment variables. The .env.example file is available in the root as a template. The root .env file is never placed under version control.

Once that's done, run the release script and await glory:

npx env-cmd npm run lerna-release -- --local false