Skip to main content

Continuous integration (CI)

All companies and projects rely on continuous integration (CI) to ensure high quality code and to avoid regressions. Because this is such a critical piece of every developer's workflow, we wanted to support it as a first-class feature within moon, and we do just that with the moon ci command.

How it works

The ci command does all the heavy lifting necessary for effectively running jobs. It achieves this by automatically running the following steps:

  • Determines touched files by comparing the current HEAD against a base.
  • Determines all targets that need to run based on touched files.
  • Additionally runs affected targets dependencies and dependents.
  • Generates an action and dependency graph.
  • Installs the toolchain, Node.js, and npm dependencies.
  • Runs all actions within the graph using a thread pool.
  • Displays stats about all passing, failed, and invalid actions.

Configuring tasks

By default, all tasks run in CI, as you should always be building, linting, typechecking, testing, so on and so forth. However, this isn't always true, so this can be disabled on a per-task basis through the runInCI or local options.

tasks:
dev:
command: 'webpack server'
options:
runInCI: false
# Or
local: true
caution

This option must be set to false for tasks that spawn a long-running or never-ending process, like HTTP or development servers. To help mitigate this, tasks named dev, start, or serve are false by default. This can be easily controlled with the local setting.

Integrating

The following examples can be referenced for setting up moon and its CI workflow in popular providers. For GitHub, we're using our setup-toolchain action to install moon. For other providers, we assume moon is an npm dependency and must be installed with Node.js.

.github/workflows/ci.yml
name: 'Pipeline'
on:
push:
branches:
- 'master'
pull_request:
jobs:
ci:
name: 'CI'
runs-on: 'ubuntu-latest'
steps:
- uses: 'actions/checkout@v4'
with:
fetch-depth: 0
- uses: 'moonrepo/setup-toolchain@v0'
- run: 'moon ci'

Choosing targetsv1.14.0

By default moon ci will run all tasks from all projects that are affected by touched files and have the runInCI task option enabled. This is a great catch-all solution, but may not vibe with your workflow or requirements.

If you'd prefer more control, you can pass a list of targets to moon ci, instead of moon attempting to detect them. When providing targets, moon ci will still only run them if affected by touched files, but will still filter with the runInCI option.

# Run all builds
$ moon ci :build

# In another job, run tests
$ moon ci :test :lint

Comparing revisions

By default the command will attempt to detect the base and head revisions automatically based on the current CI provider (powered by the ci_env Rust crate). If nothing was detected, this will fallback to the configured vcs.defaultBranch for the base revision, and HEAD for the head revision.

These values can be customized with the --base and --head command line options, or the MOON_BASE and MOON_HEAD environment variables, which takes highest precedence.

$ moon ci --base <BRANCH> --head <SHA>
# Or
$ MOON_BASE=<BRANCH> MOON_HEAD=<SHA> moon ci

Parallelizing tasks

If your CI environment supports sharding across multiple jobs, then you can utilize moon's built in parallelism by passing --jobTotal and --job options. The --jobTotal option is an integer of the total number of jobs available, and --job is the current index (0 based) amongst the total.

When these options are passed, moon will only run affected targets based on the current job slice.

GitHub Actions do not support native parallelism, but it can be emulated using it's matrix.

.github/workflows/ci.yml
# ...
jobs:
ci:
# ...
strategy:
matrix:
index: [0, 1]
steps:
# ...
- run: 'moon ci --job ${{ matrix.index }} --jobTotal 2'

Your CI environment may provide environment variables for these 2 values.

Caching artifacts

When a CI pipeline reaches a certain scale, its run times increase, tasks are unnecessarily ran, and build artifacts are not shared. To combat this, we support remote caching, a mechanism where we store build artifacts in the cloud, and sync these artifacts to machines on demand.

info

Remote caching is powered by our moonbase service. Start using today for free!

Manual persistence

If you'd prefer to not use remote caching at this time, you can cache artifacts yourself, by persisting the .moon/cache/{hashes,outputs} directories. All other files and folders in .moon/cache should not be persisted, as they are not safe/portable across machines.

However, because tasks can generate a different hash each run, you'll need to manually invalidate your cache. Blindly storing the hashes and outputs directories without a mechanism to invalidate will simply not work, as the contents will drastically change between CI runs. This is the primary reason why the remote caching service exists.

Reporting run results

If you're using GitHub Actions as your CI provider, we suggest using our moonrepo/run-report-action. This action will report the results of a moon ci run to a pull request as a comment and workflow summary.

.github/workflows/ci.yml
# ...
jobs:
ci:
name: 'CI'
runs-on: 'ubuntu-latest'
steps:
# ...
- run: 'moon ci'
- uses: 'moonrepo/run-report-action@v1'
if: success() || failure()
with:
access-token: ${{ secrets.GITHUB_TOKEN }}

The report looks something like the following: