[sabinem]

The protocol v0.12.2 was released with, step by step and by hand, since this branch predates the hotfix recipes proposed for main. Includes a fresh local stack for switching between main and the release branch, and what went wrong on the way. RELEASING.md's old advice to branch from the tag now points here instead.

[marftn]

I know this is not the final version yet, but here are some comments if that can help:

• I find the text very complicated to understand even though the topic is not that hard. It's often the case with AI so I suggest making it as concise as possible with simple sentences.
• The text references specific versions. I understand why the AI did that, but for a HOTFIX.md I think it is better to have something more generic. Providing concrete examples is good, but it would be better to use simple version examples (e.g. 0.1.1 with a hotfixes which makes it 0.1.2) rather than what the text currently has (0.12.0, then suddenly 0.12.2 and 0.12.3). Same comment for the chart version.
• I would completely drop the "Deploy to prod" section, or at least make it very generic like a hint saying "don't forget to update your current deployments". Mentioning sdsc-ordes/cloud-infra (a private repo) in a public repo is bad and makes it specific to our deployment, not to the hackathon platform in general. Also, GitOps is not enforced by this repo. Each user is free to deploy the hackathon platform however they want.
• In the same spirit, I would avoid sentences like "This branch, release/0.12, is what prod runs." This is specific to the current setup and doesn't belong in this file which is supposed to describe a protocol. A branch does not need to be running in prod to get a hotfix: any release that we still maintain should get a hotfix, whether it is deployed or not. Which versions we consider as "maintained" should probably figure in our README.

Otherwise, I think I agree with the content and think it works well with the current setup.