/jilu

Generate a change log based on the state of your Git repository.

Primary LanguageRust

记录

jìlù, to memorize

jilu generates a change log based on the state of your Git repository.


convert conventional commits into a human readable change log
––––––––––
use Git tags to annotate your releases with release titles and richly formatted release notes
–––––
customize your change log template to best serve your community
––––––––––
integrate the jilu binary into your CI workflow for automated updates



The way you structure and document your projects is personal, and transforms over time into what works best for you and your community. Jilu tries to deliver a nice and simple way to expose your project changes to the outside world. It's flexible, but it can never replace all possible workflows, and it doesn't pretend to.

Using Jilu requires using conventional commits, and while some like the rigor of structured commits, others don't, both have valid reasons. If you have an existing workflow that works for you, or an existing project without conventional commits, there won't be much to gain for you here.

What matters most is the success of your project, with or without the help of Jilu.

Quick Start

  1. Check out the change log of this project.

  2. Install jilu:

    • Using release binaries:

      curl -L https://github.com/rustic-games/jilu/releases/download/v0.3.0/jilu_0.3.0_$(uname)_x86_64.tar.gz | tar -xvf - jilu
    • Using Homebrew (soon):

      brew install jilu
    • Using Cargo:

      cargo install jilu --git https://github.com/rustic-games/jilu
  3. Visit any local repository:

    cd /path/to/repository
  4. Print change log:

    jilu
  5. Get more details (soon):

    jilu --help
  6. Integrate into your CI workflow.

About

As a fan of Conventional Commits, and an avid reader of open-source change logs, It always saddens me when a change log is missing, or is lacking important contextual details. Things like release dates, unreleased changes, or breaking changes are an important part of the public documentation of a project. On top of that, as an open-source contributor, I can't ignore the warm fuzzy feeling I get when people thank me for my contributions, but keeping track of all those contributions cuts into the time you have available for a project (still, automated thank-you's aren't as personal as an individual one, remember that).

Conventional commits and auto-generated change logs are quite popular in the JavaScript community, and there are tons of tools to help you adhere to the standards, but all the ones I tried came with downsides, the biggest ones being the lack of a single binary, lack of easy configuration, riddled with Emoji (I don't mind an Emoji or two, mind you ✌️), or trying to be a complete release management tool, which inevitably means heavy focus on the JavaScript ecosystem (and not adhering to the Unix philosophy of small single-purpose tools).

Jilu is the tool I envisioned I would like to use to generate my project change logs, and I've open-sourced it so that it may help you in your open-source endeavors as well. Have fun using it, and feel free to propose new features, or provide bug fixes!

Usage

Pipe the output of jilu to your change log file:

jilu > CHANGELOG.md

Note, if you've added custom configurations to your change log, this won't work, as Unix will empty the CHANGELOG.md file first, before jilu can read its contents, meaning it won't be able to read any existing configuration.

To work around this, use a tool like sponge to soak up the output of jilu before redirecting it to the file:

jilu | sponge CHANGELOG.md

Design

Want to know what makes Jilu tick? Read on.

Structured

The library uses existing conventions and specifications to structure the change log. Specifically, you are expected to use SemVer for tagging releases, Conventional Commits for commit messages, and the change log itself adheres to the Keep a Changelog conventions.

Automated

Release notes are generated by parsing the Git history. Clone any repository locally, and run jilu to print the generated release notes.

Commits are parsed using the Conventional Commits format.

Git tags are used to determine which commits should be part of what release. The messages of annotated tags are used to add hand-written release notes to the release. Similar to conventional commits, the first line of the tag annotation is used as the release title, the rest as the release notes.

Any commits after the latest tagged release are added to the "unreleased" section.

work in progress If a tag annotation contains a line starting with YANKED:, it will be marked as such in the change log, with anything following that marker being used as the reason for yanking the release. Git tag annotations [can be replaced after pushing them][] (while retaining the tag date and author), with some command-line-fu.

You can optionally run jilu --release MAJOR.MINOR.PATCH if you want to generate a change log with the unreleased commits grouped in a new release. When doing this, your $EDITOR will open to provide release notes. This will also create a Git tag with the same details (unless you supply --no-tag).

The release feature is added to make sure the updated change log is part of that new release. Otherwise, opening the change log in the repository state at the point of a release tag would still show the new changes in the unreleased section.

Readable

In the spirit of the Keep a Changelog guidelines, jilu produces a human-readable change log by including the following information:

  • markdown formatting for improved readability
  • custom change log introduction paragraph
  • table of contents
  • unreleased changes at the top
  • release versions, titles and dates
  • release changes grouped by type (features, fixes, etc.)
  • manually written release notes
  • short git refs linking specific commits
  • optional thank-you's to contributors
  • optional GitHub linking to release/tag/compare views

Forgiving

Commit messages that need to be excluded can be, based on a set of rules:

  • non-conventional commit messages are ignored
  • conventional commit types (such as chore) can be excluded
  • work in progress a list of blacklisted commits can be provided
  • work in progress a root commit can be provided to ignore older commits

Configurable

Jilu has a powerful configuration system that stays out of your way when you don't need it, but allows you to automate the construction of your change log in a way that works best for your project or community.

You can:

  • set header names for grouped changes (features, fixes, etc.)
  • ignore specific commit types
  • fully customize the change log template

You can check out the bottom of this project's change log for its configuration, and the default template to see how the templating system works.

Here's a quick example of what's possible (this snippet goes at the bottom of your own CHANGELOG.md file):

<!--
Config(
  github: ( repo: "rustic-games/jilu" ),
  accept_types: ["feat", "fix", "perf"],
  type_headers: {
    "feat": "Features",
    "fix": "Bug Fixes",
    "perf": "Performance Improvements"
  }
)

Template(
# My Change Log

## Upcoming Changes

{% for change in unreleased.changes %}
- {{ change.description }} ([`{{ change.commit.short_id }}`])
{%- endfor %}
)
-->

Putting the configuration inside the change log file itself ensures that the configuration can be read by Jilu, but won't show up in the markdown rendered document and is easy to ignore in text format, since it will always be at the end of the change log. It also means you don't need to add another configuration file to your Git repository root.

The templating system uses the Tera library to provide Django-like syntax. If no template is defined, the default template is used instead.