CHANGELOG.md. Then it started to generate the
Where is now and how to start using it?
"If you're not embarrassed by the first version of your product,
you've launched too late."
The first version of any software is how the author(s) think people will use it. It's like trying to see the future of people you've never met. That's why the first version is better done than perfect. The important part is to collect feedback as soon as possible and improve based on it.
Saying that I recommended ChangelogLinker to Shopsys to manage news in their monorepo. In exchange, I got informative and clear feedback with creative ideas on how to solve it from Rostislav Vitek and Petr Heinz. Huge thanks for making this tool better belongs to you guys.
First, install this package:
composer require symplify/changelog-linker --dev
If you have more than 2 major versions, your changelog can be really long and hard to orientate. Some people want to see the news in your version 3, some are still using version 1 and need to upgrade gradually.
I got this inspiration from Symfony, where they have *own CHANGELOG per minor versions:
Good idea to keep files not huge and clear. So Now it's possible to work with each CHANGELOG file separately - just use file path as the first argument of any command:
vendor/bin/changelog-linker link # "CHANGELOG.md" is used vendor/bin/changelog-linker link CHANGELOG-2.md vendor/bin/changelog-linker link CHANGELOG-3.md
This package generates changelog from merged PRs, that are not mentioned in your
Before it looks for the highest merged PR ID and added only PRs with higher id. But what if you merge PR with number 1000, but number 990 is still opened due to longer code review?
Now it works with the
merged_at instead, so no merged PR is left behind.
This is the best command to start with when you install this package for the first time.
CHANGELOG.md can be edited, items removes, shifter above or links duplicated. To keep it fit and slim, you'd have to check this manually with every link. And believe me, if you have 500+ PRs, 50+ contributors and 30+ versions, it's not as fun as you imagine.
That's where "cleanup" rocks:
vendor/bin/changelog-linker cleanup # again, you can use the file as argument vendor/bin/changelog-linker cleanup CHANGELOG-2.md
CHANGELOG.md itself it removed 50 dead lines.
Thanks to Petr Heinz ❤️️
When you generate a
CHANGELOG.md you can use
vendor/bin/changelog-linker dump-merges --in-categories
It will assign PRs to one of 4 categories: Added, Fixed, Changed and Removed. I didn't make them up, it's a standard taken from keepachangelog.com.
How it works? It uses regex to detect keywords in the pull-request title, e.g.
Added "cleanup" command → Added
Remove dead links → Removed
Fix ID detection → Fixed
You can see all the regexes in
Peter added many keywords, but also showed me a new trick : the
-private const ADDED_PATTERN = '#(add|added|adds) #i'; +private const ADDED_PATTERN = '#\b(add(s|ed|ing)?)\b#i';
It means, that words need to be standalone. Not part of any other string.
\b(is)\b applied to
"This island is beautiful" returns
A very nice trick that made detection much more precise. Thanks, Peter!