Appearance
How To Change a Practice
Goal: Improve an existing practice.
Steps
- Search our closed PRs for prior attempts to solve the problem you are seeing. Look at the objections & resolutions in order to guide your proposal.
- Make a branch named
practices/{practice-goal}/{describe-your-change}
. - Make changes. Commit often!
- Open a pull request. Invite the default reviewers.
- Manage the conversation according to the "PR Contribution Guidelines" below.
- What was the PR result?
- approved:
- One of the maintainers will merge the PR
- rejected:
- close the PR without merging it. Describe any final reasons for rejection.
- Done!
PR Contribution Guidelines
The intent is to approve changes via Consent of those impacted. Therefore:
- A PR may only be approved if everyone who is impacted has had a chance to review. This means that all default reviewers have had a chance to weigh in. These people are representatives of everyone in the organization. It is up to each of them to check in with the people they represent.
- Only comment for paramount objections. An objection is paramount if it:
- identifies some way the proposal has a reasonable likelihood of preventing us from attaining our goal, or
- identifies a resource that the proposal would likely cost that we can't afford.
- If you can resolve an objection quickly between the two of you, do so & add a commit with the changes.
- If you don't see a quick resolution, gather a small group to try to solve the objection.
- Some common solutions:
- Agree to an experiment period - try the proposal for a period of time, then auto-rollback. Agree to the measure that indicates success or failure.
- Agree to a safety measure - add some detection to the proposal. We accept the proposal and use with the agreement that if the safety detector goes off, we will revert.
- Agree to try multiple alternatives - provide a choice point in the proposal where people can do either style, but are required to take notes about how it turned out. Agree to a date on which we will auto-rollback all alternatives, review the data, and make a new proposal.