grafana/contribute/merge-pull-request.md

170 lines
9.0 KiB
Markdown
Raw Normal View History

# Merge a pull request
When a pull request has been reviewed and approved by at least one person and all checks have passed it's time to merge the pull request.
## Who is expected to merge a pull request?
Maintainers are responsible for merging all pull requests. If a maintainer has opened a pull request the general rule is that the same maintainer merges the pull request. If a non-maintainer has opened a pull request it's suggested that one of the maintainers reviewing the pull request merges the pull request.
## Checklist/summary
The following checklist/summary should give you a quick overview of what to ask/consider before merging a pull request.
- Reviewed and approved?
- All checks passed?
- Proper pull request title?
- Add to changelog/release notes?
- Needs backporting?
## Before merge
Before actually merging a pull request there's a couple of things to take into consideration.
### Status checks
Before you can merge a pull request, it must have a review approval, and all the required status checks must pass.
### Format the pull request title
The pull request title should be formatted according to `<Area>: <Summary>` (Both "Area" and "Summary" should start with a capital letter).
Keep the summary short and understandable for the community as a whole.
All commits in a pull request are squashed when merged and the pull request title will be the default subject line of the squashed commit message. It's also used for [changelog/release notes](#include-in-changelog-and-release-notes).
**Example:**
`Docs: Change url to URL in all documentation files`
See [formatting guidelines](create-pull-request.md#formatting-guidelines) for more information.
### Assign a milestone
Several things in the Grafana release process requires at least pull requests to be in a milestone, for example [generating changelog/release notes](#include-in-changelog-and-release-notes).
This makes it easier to track what changes go into a certain release. Without this information, release managers have to go through git commits which is not an efficient process.
That being said, _you don't have to assign a milestone manually_ to a pull request.
Instead, when it is merged & closed then a bot will look for the most appropriate miletone and assign it to the pull request.
That milestone should always reflect the branch that the pull request is merged into.
For every major and minor release there is a milestone ending with `.x` (e.g. `10.0.x` for the 10.0.x releases).
Pull requests targeting `main` should use the `.x` milestone of the next minor (or major) version (you can find that version number inside the `package.json` file).
Backport pull requests should use the version of the target branch (e.g. `9.4.x` for the `v9.4.x` branch).
### Include in changelog and release notes?
At Grafana we generate the [changelog](https://github.com/grafana/grafana/blob/main/CHANGELOG.md) and [release notes](https://grafana.com/docs/grafana/latest/release-notes/) based on merged pull requests. Including changes in the changelog/release notes is very important to provide a somewhat complete picture of what changes a Grafana release actually includes.
There's a GitHub action available in the repository named [Update changelog](https://github.com/grafana/grafana/blob/main/.github/workflows/update-changelog.yml) that can manually be triggered to re-generate the changelog and release notes for any release.
Exactly what changes should be added to the changelog is hard to answer but here's some general guidance:
- Include any bug fix in general.
- Include any change that you think would be interesting for the community as a whole.
- Skip larger features divided in multiple pull requests since they might go into the release blog post/What's New article.
- Use your best judgement and/or ask other maintainers for advice.
- Including a change in error rather than skipping one that should have been there is better.
- Always keep [Format the pull request title](#format-the-pull-request-title) in mind.
An active decision to include change in changelog/release notes needs to be taken for every pull request. There's a pull request check named **Changelog Check** that will enforce this. By adding/removing labels on the pull request or updating the pull request title/description the check will be re-evaluated.
#### Skip changelog
If you don't want to include your change in changelog/release notes you need to add a label named **no-changelog** to the pull request.
#### Include in changelog/release notes
To include a pull request in the changelog/release notes you need to add a label named `add to changelog` to the pull request. Then additional validation rules is checked:
- Title need to be formatted according to [Format the pull request title](#format-the-pull-request-title)
- Description needs to include a breaking change notice if change is labeled to be a breaking change, see Breaking changes below for more information.
Not complying with above rules can make the **Changelog Check** fail with validation errors.
The changelog/release notes are divided into sections and here's a description of how you make a pull request show up in a certain section.
**Features and enhancements:**
Labeled with `add to changelog` and any of the other section rules don't apply.
**Bug fixes:**
Labeled with `add to changelog` and either labeled with `type/bug` or the pull request title contains `fix` or `fixes`.
**Plugin development fixes & changes:**
Labeled with `area/grafana/ui` or `area/grafana/runtime`.
**Deprecations:**
In case the pull request introduces a deprecation you should document this. Label the pull request with `add to changelog` and use the following template at the end of the pull request description describing the deprecation change.
```md
# Deprecation notice
<Deprecation description>
```
**Breaking changes:**
In case the pull request introduces a breaking change you should document this. Label the pull request with `add to changelog` and `breaking change` and use the following template at the end of the pull request description describing the breaking change.
```md
# Release notice breaking change
<Breaking change description>
```
### Should the pull request be backported?
Backporting is the process of copying the pull request into the version branch of one or multiple previous releases.
This should only be done for _critical bug fixes_ and involves the intervention of a Grafana Labs employee.
To make this decision explicit, there is a **Backport Check** that is re-evaluated automatically by adding/removing labels on the pull request.
#### No backport
If you don't want to backport you need to add a label named **no-backport** to the pull request.
This should be the default!
#### Backport
If your pull request fixes a critical bug and needs to go into one or several existing release branches it should be backported.
For a pull request to be backported, please add it to "Critical Bug Release" form so that the delivery team is aware that a release for a previous version needs to be made.
Once approved, the backporting process will continue from there.
In the meantime, set the `no-backport` label.
#### Required labels
To ensure that we don't backport pull requests that don't need to be backported, i.e. implement new features, and only backport pull requests that address bugs, have a product approval, or refer to docs changes, backport labels need to be followed by either:
- `type/bug` label: Pull requests which address bugs,
- `product-approved` label: Urgent fixes which need product approval, in order to get merged,
- `type/docs` label: Docs changes.
- `type/ci` label: Changes to the CI automation.
> **Note:** You can still backport a pull request after it's been merged.
## Doing the actual merge
Time to actually merge the pull request changes. All commits in a pull request are squashed, hence the GitHub `Squash and merge` button is used to initialize the merge.
This will present you with options allowing you to optionally change the commit message before merging. Please remember that developers might use the commit information when reviewing changes of files, doing git blame and resolving merge conflicts etc., trying to quickly figure out what the actual change was. But there's not really any best practices around this, the following is an attempt to bring some guidance.
Do:
- Make sure the pull request title is formatted properly before merging, this will automatically give you a good and short summary of the commit/change.
- Leave `Co-authored-by:` lines as is so that co-authors will be accounted for the contribution.
- Remove any commit information that doesn't bring any context to the change.
Consider:
- Add any references to issues that the pull request fixes/closes/references to ease giving quick context to things. Doing this allows cross-reference between the commit and referenced issue(s).
Finalize the merge by clicking on the `Confirm squash and merge` button.
## After the merge
Make sure to close any referenced/related issues. It's recommended to assign the same milestone on the issues that the pull request fixes/closes, but not required.