Releases are mostly automated using release-it and lerna-changelog.
Since the majority of the actual release process is automated, the primary
remaining task prior to releasing is confirming that all pull requests that
have been merged since the last release have been labeled with the appropriate
lerna-changelog
labels, and the titles have been updated to ensure they
represent something that would make sense to our users. Some great information
on why this is important can be found at
keepachangelog.com, but the overall
guiding principle here is that changelogs are for humans, not machines.
When reviewing merged PR's the labels to be used are:
- breaking - Used when the PR is considered a breaking change.
- enhancement - Used when the PR adds a new feature or enhancement.
- bug - Used when the PR fixes a bug included in a previous release.
- documentation - Used when the PR adds or updates documentation.
- internal - Used for internal changes that still require a mention in the changelog/release notes.
Once the prep work is completed, the actual release is straight forward:
-
First, ensure that you have an environment variable with your GitHub token setup as
GITHUB_AUTH
. This token will be used for generating your changelog (unauthenticated requests to the GitHub API are heavily throttled) and for creating the GitHub release. Only "repo" access is needed; no "admin" or other scopes are required. -
Next, ensure that you have installed your projects dependencies:
yarn install
- And last (but not least 😁) do your release:
yarn release
release-it manages the actual
release process. It will prompt you to choose the version number after which
you will have the chance to hand tweak the changelog to be used (for the
CHANGELOG.md
and GitHub release), then release-it
continues on to tagging,
pushing the tag and commits, etc. Finally, GitHub Actions will build the commit
and push the release to npm.