Discuss about a possible release-changelog action

[jorgepiloto]

Hi all, I wanted to get your feedback on a design decision about an automated changelog for our libraries.

Background 📝
The idea is to use https://towncrier.readthedocs.io/en/stable/. This tool is based on the idea of "fragments". Everytime you make a change in your code you need to write a small RST file. This file is named .. For example:

1234.bugfix
2345.feature

When you are ready to release, you can execute "towncrier build ". The tool glues all the fragments and builds a new entry for the changelog.

The issue 🐞
Let's say that a developer performs a release using an action named release-changelog. The actions runs the towncrier build ... command and the desired changelog file gets updated.

These new changes are new, so a new tag is required.

This introduces a chicken-egg problem: tag -> update CHANGELOG -> new changes -> commit changes -> tag

Possible solutions 💡

1. Create a changelog-style action that checks that the CHANGELOG contains the tag that triggered the workflow. This forces developers to run locally the towncrier build ... action, manually commit the updated changelog, pushing and finally tagging before the release.

[RobPasMue]

It is not possible that we are the first to experience this issue... I'm looking at their docs and I might fail to see the problem here. I'd appreciate some kind of pair-programming session @jorgepiloto - I'd be happy to help.