Re: [che-dev] Conventional commits

[Thomas Mäder <tmader@xxxxxxxxxx>]

You didn't think I would not have an opinion, did you ;-) 

tl;dr: I don't think we should neither enforce nor encourage conventional commits. Here's why:

1. It's aimed at the wrong level

The stated goal of this effort is to make it easier for management/architects to follow the progress of coding by looking at which PR's have been merged. So if we want to improve something, we need to improve the PR titles, not the commit messages.  Commit messages are communicating to programmers: when I look at a line of code in a year's time, I want to understand the why of that line of code. Mario, if you as an architect are reading my commit messages, you're acting at the wrong level of abstraction. That's what you have team leads for, if anything.

2. It doesn't help

I have recently seen a conversation on an issue about whether something is a "fix" or a "chore". I fail to understand how getting this terminology correct will improve the product. It's the same useless discussion as whether something is a "regression", a "bug" or a "feature". I really don't care: does it need doing or not is the question.

3. It's culturally wrong
Focusing on this kind of formal correctness does not foster a culture of ownership and responsibility. On the contrary, it absolves people from the responsibility of writing meaningful commit messages. It's the "but the test suite passed, so it's no my fault"-mindset. Great software comes from great developers doing the right thing, not from process.

4. It hides important information

Here's an example from a recent commit on che-theia as seen in the github UI:

"fix(convert): When converting devfile v1 to v2, registryUrl attribute… "

The continuation of the message is hidden, it is:

"… is forgotten"

The message without the "semantic commit" stuff would read:

"When converting devfile v1 to v2, registryUrl attribute is forgotten"

Now that actually makes sense, right?

/Thomas

------ Original Message ------

From: "Mario Loriedo" <mario.loriedo@xxxxxxxxx>

To: "che developer discussions" <che-dev@xxxxxxxxxxx>

Sent: 15/06/2021 18:41:45

Subject: [che-dev] Conventional commits

As mentioned during yesterday's community call we should have a discussion about conventional commits [1] to decide if we should enforce it or not.

Currently a PR check verifies if commits messages follow the convention. PRs can be merged even if the check fails. But we need to decide if we are going to enforce the convention (block PRs where the check fails, update the contribution guide etc...) or remove the PR check and let contributors free to use any commit message format they like.

What's your thoughts?

[1] https://www.conventionalcommits.org/