Contributing
You want to start contributing? Nice!
We do use Gerrit for Code Review. It allows a more granular review (per-commit granularity rather than PR granularity), as well as keeping track as how commits change over time. It greatly simplifies the review process, and leads to overall more high-quality contributions.
While it might initially look a bit intimidating, you hopefully will spend less time learning its workflow than writing actual Snix code.
This assumes you have the repo already cloned and the necessary tools installed as described in Building Snix , so make sure you went through these instructions first.
Creating a Gerrit account
- Navigate to our Gerrit instance . Hit the “Sign in” button (which allows SSO with some common IdPs)
- In the User settings, paste an SSH public key and hit the “Add New SSH key” button. 1
- Alternatively, you can also create “HTTP Credentials” (though saving the HTTP password is messy).
Update your git remote URL
Instead of trying to push to Forgejo, reconfigure your git remote URL to interact with Gerrit directly.
Replace $USER
with your Username
shown in the Gerrit settings.
If using SSH authentication:
$ git remote set-url origin "ssh://$USER@cl.snix.dev:29418/snix"
If using HTTP authentication:
$ git remote set-url origin "https://$USER@cl.snix.dev/a/snix"
Install the commit-msg hook
Gerrit uses a commit-msg
hook to add a Change-Id: …
field to each commit
message if not present already. This allows Gerrit to identify new revisions /
updates of old commits, and track them as new revisions of the same “CL” 2.
To install the commit-msg hook, run the following from the repo root:
$ mkdir -p .git/hooks
$ curl -Lo .git/hooks/commit-msg https://cl.snix.dev/tools/hooks/commit-msg
$ chmod +x .git/hooks/commit-msg
Did you know?
Gerrit refuses receiving commits without these Change-Id: …
fields.
If you already have some local commits without Change-Id
field, git commit --amend
them after installing the commit-msg
hook to add them.
Push your changes
Do some local changes, and push them to Gerrit as follows:
$ git push origin HEAD:refs/for/canon
Gerrit will print links to newly created CLs to your terminal.
If you want to update/edit your CL, simply squash these changes into your local commit and push again.
The Gerrit model
If do not have experience with the Gerrit, consider reading the Working with Gerrit: An example or Basic Gerrit Walkthrough — For GitHub Users .
Some more tips:
- Assign a reviewer to review your changes.
- React on comments and mark them as resolved once you did.
- Comments are only “Drafts” (stored server-side) until you send them off.
This can be done by the
Reply
button on the top, for example.
- Comments are only “Drafts” (stored server-side) until you send them off.
This can be done by the
- Once CI is green, it’s up to the Author of the CL to submit, not the
reviewer.
If you want a bot to automatically submit in this case, you can add the
Autosubmit+1
label. - Rebase on
origin/canon
regularly. You cannot push if you still have an old version of a now-submitted CL in your git log.
Did you know?
You can immediately assign reviewers and other fields while pushing a new/updated change, by adding it to the push URL.
$ git push origin HEAD:refs/for/canon%r=alice,cc=bob,l=Autosubmit+1,publish-comments
- will set
alice
as a reviewer - will set
bob
as CC - adds the
Autosubmit+1
label - publishes any outstanding draft comments
Commit conventions
Commit messages
The following described way of writing commit messages is known as Conventional Commits and should bring these advantages:
- automatic creation of changelogs from commit messages
- lean commits: one focused change per commit
- filtering of commit history by type
All commit messages should be structured like this:
type(scope): Subject line with at most a 72 character length
Body of the commit message with an empty line between subject and
body. This text should explain what the change does and why it has
been made, *especially* if it introduces a new feature.
Relevant issues should be mentioned if they exist.
Where type
can be one of:
feat
: A new feature has been introducedfix
: An issue of some kind has been fixeddocs
: Documentation or comments have been updatedstyle
: Formatting changes onlyrefactor
: Hopefully self-explanatory!test
: Added missing tests / fixed testschore
: Maintenance work
And scope
should refer to some kind of logical grouping inside of the project.
It does not make sense to include the full path unless it aids in
disambiguating. For example, when changing /snix/eval/src/lib.rs
it is enough to write feat(eval): ...
.
Please take a look at the existing commit history for examples.
Commit content
Multiple changes should be divided into multiple git commits whenever possible. Common sense applies.
Code quality
This one should go without saying - but please ensure that your code quality does not fall below the rest of the project. This is of course very subjective, but as an example if you place code that throws away errors into a block in which errors are handled properly your change might be rejected.