- 1 Before You Begin
- 2 Your first commit
- 3 Reviews
- 4 Testing
Before You Begin
To familiarize yourself with Magnum, try it out using the information in our Dev Quickstart Guide. When you are ready to start contributing, you will need to execute an OpenStack CLA. This is required before you can submit reviews to our Magnum Repo. For information about how prepare for contribution, please consult the How To Contribute wiki page.
Learn About Gerrit
Be sure to read the Developer's Guide for information about how to submit your commit for review so it can be merged into the Magnum code base.
Setting up your git review settings
git config --global user.name "Firstname Lastname" git config --global user.email "email@example.com" git config --global gitreview.username "your_launchpad_username"
To check your git configuration:
git config --list
On Ubuntu, MacOSx, or most other Unix-like systems, it is as simple as:
pip install git-review
There are other installation options detailed in the Developer's Guide. You can now check out the Magnum code and begin working on it:
Your first commit
Set up your local branch
git clone git://git.openstack.org/stackforge/magnum cd magnum git checkout -b [branch name] git review -s
Write some awesome code
At this point can write your code and push it to Gerrit for review.
git add <list of files you added/changed> git commit -a git review -v --draft
Once you are happy with your code and want it to be reviewed you want to convert it from a Draft. "Sign In" at https://review.openstack.org/ and after verifying the review yourself hit the "Publish" button on the page.
If you know you are ready for others to review your code, you can skip the draft step and just do:
git review -v
If you want to revise your patchset in the review system in response to feedback, make your changes, then use:
git commit -a --amend git review -v
Upon approval of the review your code will be automatically merged.
The OpenStack CI system uses the concept of core reviewers. These are individuals who have consistently reviewed code for the project, and helped over a considerable period of time to improve the quality and consistency of what we merge into the code base. Project contributors will feel that this reviewer is a positive influence on the team and that they maintain the values and traditions of the OpenStack development community.
Existing core reviewers may nominate new ones in an ML thread. Consent among the current reviewers shall result in the declaration of the new core reviewer by the PTL. Lack of unanimous consent shall be carefully considered, and a final decision informed by input from from active team members shall be made by the PTL. Core reviewers who are judged by their peers in the core review group to fall short of the expectations for contribution of a core reviewer may be nominated for return to regular reviewer status.
Patches require a core reviewer to mark a review as "Approved" before they are merged.
Code Approval for Merge
- For Approval, two core reviewers shall supply a
Continuing Someone Else's Contribution
- If a patch submitted by one contributor is picked up and completed by another contributor, special handling of the resolution should be used.
Advice for Reviewers
-1vote is an opportunity to make our code better before it is merged. Please do your best to make helpful, actionable -1 votes.
- Avoid the temptation to blindly
+1code without reviewing it in sufficient detail to form an opinion.
- When voting
-1on a patch, it means that you want the submitter to make a revision in accordance with your feedback before core reviewers should consider this code for merge.
- If you ask a question, you should vote
0unless you anticipate that the answer to that question is likely to cause you to vote against the patch without further revisions.
- If you use a
-1vote for a question, and the contributor answers the question, please respond acknowledging the question. Either change your vote or follow up with additional rationale for why this should remain a
-2vote is a veto by a single core reviewer. It is sticky. That means that even if you revise your patch, that vote will persist. To allow your patch to merge, that same reviewer must clear the
-2vote first. This vote is used when you have contributed something that is not in alignment with the current project vision, or is implemented in a way that can not be accepted. For example, security concerns that a core reviewer wants to individually re-evaluate before allowing the contribution to continue. It can also be used as a way to halt further gate testing of a patch, if something is included that may break the gate. It works even after a
2*+2,+Aapproval for merge, but before the patch reaches MERGED state.
- To avoid a
-2vote, discuss your plans with the development team prior to writing code, and post a WIP (workflow-1) patch while you are working on it, and ask for input before you submit it for merge review.
See our Magnum/Testing wiki.