Skip to content

Latest commit

 

History

History
378 lines (259 loc) · 12.9 KB

README.md

File metadata and controls

378 lines (259 loc) · 12.9 KB
Raw

Standard Version

Build Status NPM version Coverage Status Conventional Commits community slack

Having problems? want to contribute? join our community slack.

Automate versioning and CHANGELOG generation, with semver and conventional commit messages.

how it works:

  1. when you land commits on your master branch, select the Squash and Merge option.
  2. add a title and body that follows the Conventional Commits Specification.
  3. when you're ready to release:
  4. git checkout master; git pull origin master
  5. run standard-version
  6. git push --follow-tags origin master && npm publish (or, docker push, gem push, etc.)

standard-version does the following:

  1. bumps the version in metadata files (package.json, composer.json, etc).
  2. uses conventional-changelog to update CHANGELOG.md
  3. commits package.json (et al.) and CHANGELOG.md
  4. tags a new release

Installation

As npm run script

Install and add to devDependencies:

npm i --save-dev standard-version

Add an npm run script to your package.json:

{
  "scripts": {
    "release": "standard-version"
  }
}

Now you can use npm run release in place of npm version.

This has the benefit of making your repo/package more portable, so that other developers can cut releases without having to globally install standard-version on their machine.

As global bin

Install globally (add to your PATH):

npm i -g standard-version

Now you can use standard-version in place of npm version.

This has the benefit of allowing you to use standard-version on any repo/package without adding a dev dependency to each one.

Configuration

You can configure standard-version either by:

  1. Placing a standard-version stanza in your package.json (assuming your project is JavaScript).
  2. Creating a .versionrc, .versionrc.json or .versionrc.js.
  • If you are using a .versionrc.js your default export must be a configuration object, or a function returning a configuration object.

Any of the command line parameters accepted by standard-version can instead be provided via configuration. Please refer to the conventional-changelog-config-spec for details on available configuration options.

Customizing CHANGELOG Generation

By default, standard-version uses the conventionalcommits preset.

This preset:

  • adheres closely to the conventionalcommits.org specification.
  • is highly configurable, following the configuration specification maintained here.
    • we've documented these config settings as a recommendation to other tooling makers.

There are a variety of dials and knobs you can turn related to CHANGELOG generation.

As an example, suppose you're using GitLab, rather than GitHub, you might modify the following variables:

  • commitUrlFormat: the URL format of commit SHAs detected in commit messages.
  • compareUrlFormat: the URL format used to compare two tags.
  • issueUrlFormat: the URL format used to link to issues.

Making these URLs match GitLab's format, rather than GitHub's.

CLI Usage

NOTE: To pass nested configurations to the CLI without defining them in the package.json use dot notation as the parameters e.g. --skip.changelog.

First Release

To generate your changelog for your first release, simply do:

# npm run script
npm run release -- --first-release
# or global bin
standard-version --first-release

This will tag a release without bumping the version in package.json (et al.).

When ready, push the git tag and npm publish your first release. \o/

Cut a Release

If you typically use npm version to cut a new release, do this instead:

# npm run script
npm run release
# or global bin
standard-version

As long as your git commit messages are conventional and accurate, you no longer need to specify the semver type - and you get CHANGELOG generation for free! \o/

After you cut a release, you can push the new git tag and npm publish (or npm publish --tag next) when you're ready.

Release as a pre-release

Use the flag --prerelease to generate pre-releases:

Suppose the last version of your code is 1.0.0, and your code to be committed has patched changes. Run:

# npm run script
npm run release -- --prerelease

you will get version 1.0.1-0.

If you want to name the pre-release, you specify the name via --prerelease <name>.

For example, suppose your pre-release should contain the alpha prefix:

# npm run script
npm run release -- --prerelease alpha

this will tag the version 1.0.1-alpha.0

Release as a target type imperatively like npm version

To forgo the automated version bump use --release-as with the argument major, minor or patch:

Suppose the last version of your code is 1.0.0, you've only landed fix: commits, but you would like your next release to be a minor. Simply do:

# npm run script
npm run release -- --release-as minor
# Or
npm run release -- --release-as 1.1.0

you will get version 1.1.0 rather than the auto generated version 1.0.1.

NOTE: you can combine --release-as and --prerelease to generate a release. This is useful when publishing experimental feature(s).

Prevent Git Hooks

If you use git hooks, like pre-commit, to test your code before committing, you can prevent hooks from being verified during the commit step by passing the --no-verify option:

# npm run script
npm run release -- --no-verify
# or global bin
standard-version --no-verify

Signing commits and tags

If you have your GPG key set up, add the --sign or -s flag to your standard-version command.

Lifecycle scripts

standard-version supports lifecycle scripts. These allow you to execute your own supplementary commands during the release. The following hooks are available and execute in the order documented:

  • prerelease: executed before anything happens. If the prerelease script returns a non-zero exit code, versioning will be aborted, but it has no other effect on the process.
  • prebump/postbump: executed before and after the version is bumped. If the prebump script returns a version #, it will be used rather than the version calculated by standard-version.
  • prechangelog/postchangelog: executes before and after the CHANGELOG is generated.
  • precommit/postcommit: called before and after the commit step.
  • pretag/posttag: called before and after the tagging step.

Simply add the following to your package.json to configure lifecycle scripts:

{
  "standard-version": {
    "scripts": {
      "prebump": "echo 9.9.9"
    }
  }
}

As an example to change from using GitHub to track your items to using your projects Jira use a postchangelog script to replace the url fragment containing 'https://github.com/`myproject`/issues/' with a link to your Jira - assuming you have already installed replace

{
  "standard-version": {
    "scripts": {
      "postchangelog": "replace 'https://github.com/myproject/issues/' 'https://myjira/browse/' CHANGELOG.md"
    }
  }
}

Skipping lifecycle steps

You can skip any of the lifecycle steps (bump, changelog, commit, tag), by adding the following to your package.json:

{
  "standard-version": {
    "skip": {
      "changelog": true
    }
  }
}

Committing generated artifacts in the release commit

If you want to commit generated artifacts in the release commit (e.g. #96), you can use the --commit-all or -a flag. You will need to stage the artifacts you want to commit, so your release command could look like this:

"prerelease": "webpack -p --bail",
"release": "git add <file(s) to commit> && standard-version -a"

Dry run mode

running standard-version with the flag --dry-run allows you to see what commands would be run, without committing to git or updating files.

# npm run script
npm run release -- --dry-run
# or global bin
standard-version --dry-run

Prefix Tags

Tags are prefixed with v by default. If you would like to prefix your tags with something else, you can do so with the -t flag.

standard-version -t @scope/package\@

This will prefix your tags to look something like @scope/package@2.0.0

If you do not want to have any tag prefix you can use the -t flag without value.

CLI Help

# npm run script
npm run release -- --help
# or global bin
standard-version --help

Code usage

Use the silent option to stop standard-version from printing anything to the console.

const standardVersion = require('standard-version')

// Options are the same as command line, except camelCase
// standardVersion returns a Promise
standardVersion({
  noVerify: true,
  infile: 'docs/CHANGELOG.md',
  silent: true
}).then(() => {
  // standard-version is done
}).catch(err => {
    console.error(`standard-version failed with message: ${err.message}`)
})

Commit Message Convention, at a Glance

patches:

git commit -a -m "fix(parsing): fixed a bug in our parser"

features:

git commit -a -m "feat(parser): we now have a parser \o/"

breaking changes:

git commit -a -m "feat(new-parser): introduces a new parsing library
BREAKING CHANGE: new library does not support foo-construct"

other changes:

You decide, e.g., docs, chore, etc.

git commit -a -m "docs: fixed up the docs a bit"

but wait, there's more!

Github usernames (@bcoe) and issue references (#133) will be swapped out for the appropriate URLs in your CHANGELOG.

Badges!

Tell your users that you adhere to the Conventional Commits specification:

[![Conventional Commits](https://img.shields.io/badge/Conventional%20Commits-1.0.0-yellow.svg)](https://conventionalcommits.org)

FAQ

How is standard-version different from semantic-release?

semantic-release is described as:

semantic-release automates the whole package release workflow including: determining the next version number, generating the release notes and publishing the package.

While both are based on the same foundation of structured commit messages, standard-version takes a different approach by handling versioning, changelog generation, and git tagging for you without automatic pushing (to GitHub) or publishing (to an npm registry). Use of standard-version only affects your local git repo - it doesn't affect remote resources at all. After you run standard-version, you can review your release state, correct mistakes and follow the release strategy that makes the most sense for your codebase.

We think they are both fantastic tools, and we encourage folks to use semantic-release instead of standard-version if it makes sense for their use-case.

Should I always squash commits when merging PRs?

The instructions to squash commits when merging pull requests assumes that one PR equals, at most, one feature or fix.

If you have multiple features or fixes landing in a single PR and each commit uses a structured message, then you can do a standard merge when accepting the PR. This will preserve the commit history from your branch after the merge.

Although this will allow each commit to be included as separate entries in your CHANGELOG, the entries will not be able to reference the PR that pulled the changes in because the preserved commit messages do not include the PR number.

For this reason, we recommend keeping the scope of each PR to one general feature or fix. In practice, this allows you to use unstructured commit messages when committing each little change and then squash them into a single commit with a structured message (referencing the PR number) once they have been reviewed and accepted.

License

ISC