Evaluating the change from semantic-release to changelogen

[amtins]

As a Pillarbox developer, I'd like to know if changelogen can be used instead of semantic-release to potentially remove a number of development dependencies.

Acceptance Criteria

• use changelogen
• remove all semantic-release dependencies

[jboix]

I've made a test in a forked project, it seems that is working fine with no configuration, see the Release that it generated.

By default it also generates and commits a CHANGELOG in the project, the changelog can be disabled by adding the option --no-output if desired.

I've submitted a draft based on the forked project: #172

[amtins]

@jboix the issue that I have is how the SemVer is handled and its consistency, a few examples:

• Git test only ! is used to create a breaking change version, you can't have a body with BREAKING CHANGE
• If the version is less than 1.x.x the breaking change doesn't seem to work well
• It's also strange that, let's say the last hash commit is cafebabe which is available on HEAD and on the latest tag. Running npx changelogen@latest --release shouldn't create a new version if nothing has changed. However it generates a new version with a new tag. GetGitDiff seems to behave as expected and returns an empty array when there is no diffs. In the code I see multiple issues starting from https://github.com/unjs/changelogen/blob/382eda92d13918fa5e5a354eaac04592a3fb873d/src/commands/default.ts#L44, https://github.com/unjs/changelogen/blob/382eda92d13918fa5e5a354eaac04592a3fb873d/src/git.ts#L112

That said, I absolutely love this project and how easy it is to configure and I would love to use it. Maybe we could help them fixing these minor issues. What do you think?

[amtins]

Thank you @jboix for the PR handling the BREAKING CHANGE https://github.com/unjs/changelogen/pull/168 👍🏽 👌🏽

[amtins]

We've decided to keep semantic-release as it offers more functionalities.