To contribute code to Matomo you will need to create a pull request. Below we explain how the Matomo core team works with pull requests.
This is the overall process for getting a PR merged from PR creation to merging a PR. For more details read the entire page.
Draft(you can do this in the
Reviewerssection). Once the PR is done and ready for a review click on
Ready for reviewand add the label
Needs Review. If a PR references another issue we assign the label
not-in-changelog. The PR author also assigns the PR to a milestone (usually the same milestone as the referenced issue).
Needs Reviewwhen changes are required. No new label is added for now for simplicity
Needs Reviewagain once it's ready for another review.
Now that your changes are published, you can send them in a pull request to the main Matomo project.
To do so, visit your fork on GitHub and select the bugfix branch in the branch selector (located right above the directory listing on the left side of the page). Then click the Pull Request button in the middle of the top of the page.
On this screen you'll be able to see exactly what changes you've made by looking at your commits and at what files have been changed. You can use this as an opportunity to review your changes.
Read Creating Pull requests: best practises to maximise your changes to create a great pull request.
Once you're ready to create the pull request, write a description of the pull request and any notes that are important for the person who will review your code, and then click Send pull request. This will create a new pull request which you will be able to find and view here. The description should contain as much as necessary and as little as possible.
If the PR is ready for a review, assign the label
Needs Review and put it in the correct milestone. The milestone for the PR is usually the same milestone as the original issue you worked on. If there is an issue for this PR, then we also assign the label
Not In Changelog. This prevents the same issue being listed twice in the changelog. A PR will only be reviewed when it has the
Needs Review label.
If the PR is not ready for a review yet and the PR is in progress, then you can click on the link
Convert to draft which you find in the Github PR UI below
Reviewers. Once you finished the work for the PR and it's ready for a review, you can click on
Ready for review where it says
This pull request is still a work in progress and assign the labels as mentioned above.
For more PR best practices read below.
And if you are a core developer, if there is an issue that the pull request will fix, and the issue is in a milestone, set the milestone of the pull request to the same one as the issue.
Once your pull request is public, developers will review it and leave comments regarding what should be changed. You can then make changes, commit them and then push them to your remote repository (origin) and they will automatically be shown in the pull request.
If you find yourself building a simple test page to test a certain feature (like a tracking feature or a certain widget), then it may be useful to contribute this examples to our test examples repository so other people can reuse such a page.
Here are best practises we aim to follow when creating, reviewing and merging pull requests:
In the ideal efficiency case, a Pull Request can be approved immediately. Obviously this is not always possible and many times useful information, considerations, edge cases, etc will be added by the reviewer which is also valuable. However, in aiming for fewer communication round trips during reviews, the following points should be considered.
fix #11111) in the PR description. This way, the related issue will be closed automatically when the PR is merged.
As core developers one of our primary responsibilities is to review and merge other pull requests. This document lays out the general process and things to look for. In the pull request template on github, there is a checklist of reminders of what to look for. Here, we'll go into details:
For every pull request it is expected that the reviewer will actually check out the code locally and test it. We want to make sure it does the thing it's supposed to, and handles any error conditions gracefully.
This means manually testing and looking for possible issues in the submitted code. It's required for pull request changes to have a visible effect on the tests (in most cases), but we don't want to rely on them alone, since it's always a possibility for people to make mistakes. The review is a chance to catch them before they become bugs.
Any problems found in the logic of a change should ideally result in new tests.
When manual testing it's also required to think about any edge cases that might occur and cause issues. Running Matomo in the cloud, we've learned that edge cases, despite their name, do occur every now and then, and cause problems. It's far better to try and avoid these problems before we merge PRs.
Some starting points that could help when looking for edge cases:
By nature, edge cases are hard to find, but it's definitely better to catch them beforehand, rather than have to debug cloud, or worse, debug a user's Matomo via email.
If a feature touches Matomo's UX in any way, Thomas and Matt are required to provide their input (ping them if needed). But, we also want developers to think about usability themselves.
When reviewing a pull request, think about whether it's possible that users may be confused by how it works, or unsure of how to use it. If there's a possibility that they might reach out to support or the forums or make a github issue, then we'd like to prevent that, either with a change to the pull request, or by creating/editing a faq so we can address these requests quickly.
en.jsonlanguage file. We don't put any translations into any other language files as these are provided by the translators to ensure consistent wording is used and to ensure the translations were reviewed.
We also want to make sure there are no security issues introduced by this pull request. We've created a checklist here for some security issues to look out for: https://developer.matomo.org/guides/security-in-piwik#checklist, but there are many other ways vulnerabilities can manifest. And we very much want to prevent any from getting into the codebase.
The code review is just that, looking for mistakes in the code, along with ways the code could be clearer, just in case the pull request author missed something. If you see something that could be done with less code, or see something that confuses you as another developer, please bring it up.
We also find that most review items are best stated as questions, and not as demands, in order to foster a more positive environment that values collaboration over argument.
We want to take advantage of the benefits of automated testing as much as possible. If a pull request can be tested, it should be. OR it should at least show a change in the existing tests. This also helps to prove the feature or fix does what it's supposed to.
The level of testing would vary based on what is being reviewed, but some form of test is required, unless it is really not possible.
If a change touches something users actively use or a piece of code that is considered public API for plugin developers or those integrating Matomo, then we want to make sure our change doesn't break anything these users and developers might currently be doing.
This is fairly simple for code (for example, if we add a new parameter to a function considered public API we want to make sure it has a default value, so people currently calling it won't encounter an error after updating Matomo). It's, unfortunately, a bit more complicated with users. There are many ways users manage to use Matomo, and keeping things working the way they currently want it can be a challenge.
We want to make sure API methods still behave as they did before for the same inputs, old links still go to the same pages, CLI commands do not fail because parameters were removed, and many other things. It can be hard to consider given there are so many ways we can break something for a user, but it's definitely important to keep our users workflows working.
There are many ways we might break the PHP API. For example:
The developer changelog is located at https://github.com/matomo-org/matomo/blob/4.x-dev/CHANGELOG.md. If a change affects the work of plugin developers or developers who integrate Matomo into their websites, we'd want to mention the change in the developer changelog. This can include:
A change that will force them to make modifications to their plugins or integrations. As said in a previous section, we like to avoid situations like this (except for major releases), but sometimes it's unavoidable.
new features, API methods, configuration options
If a new feature is something developers may want to take advantage of, then we want to mention it so more people become aware of it.
changes to existing features
If we change how an existing feature or API works, we want to mention it to developers. These are not breaking changes, those are mentioned above, just changes in the how something works. An example would be when we introduced an allowlist for trusted hosts to download geoip databases from. It's unlikely this would break anything for existing users, but it's still worth mentioning.
If we deprecate some code or API, we want to mention it as early as possible, so developers have ample time to stop using it before we remove it.
If the change is for a new feature or affects the way an existing feature works, then we'll want to modify the existing documentation (or create new documentation if it doesn't exist). For changes that affect plugin developers and developers integrating Matomo, we'd want to document the changes within our existing phpdocs and the developer documentation website: https://github.com/matomo-org/developer-documentation
Changes to features should be reflected in changes to user documentation. For new features we may give the task of writing documentation to technical writers or our support team, but for smaller changes, developers might be expected to make those themselves.
We should also think about whether new faqs should be created or if we need to modify existing ones.
If a screenshot changes significantly, then we should also update screenshots on our website. If only a wording changes, then usually it's not needed to update the screenshot. There are no hard rules when or when not to update a screenshot. If basically depends if the screenshot is still clear in the context of the content.
External contributor pull requests should be reviewed in the same way as PRs from core developers, EXCEPT:
we should assume that they may not want to or have the time to completely fix up their pull requests. It's possible we core devs may need to fix the build, or apply some changes, or other minor things. We shouldn't just take over the PR unless it is an easy task or a very useful thing to have merged.
we should assume they don't have the technical knowledge we have of Matomo and may need more help than a core dev would.
and we should always thank them for the contribution! It's always a good sign when people take an interest in our product, and it's pretty amazing when people decide to work for free :)
Say "we" instead of "you". Eg Can we rename this variable to something more descriptive, like seconds_remaining? vs You misspelled 'successfully.' It minimizes the risk of raising your team mate's defences
Frame feedback as requests, not commands
Tie notes to principles, not opinions, when possible. Eg Instead of saying, "We should split this class into two," it's better to say, "Right now, this class is responsible for both downloading the file and parsing it. We should split it up into a downloader class and parsing class per the single responsibility principle." You can't always articulate exactly what is wrong with a piece of code in terms of established principles. Sometimes code is just ugly or unintuitive, and it's hard to pin down why. In these cases, explain what you can, but keep it objective. If you say, "I found this hard to understand," that's at least an objective statement, as opposed to, "this is confusing," which is a value judgement and may not be true for every person.
Limit feedback on repeated patterns. When you notice that several of the author's mistakes fit the same pattern, don't flag every single instance. Point it out maybe 2 or 3 times and then rather point out the pattern not each individual mistake.
Offer sincere praise: Reviews are a valuable opportunity to reinforce positive behaviors. Any time you see something in the change list that delights you, tell the author about it:
Grant approval when remaining fixes are trivial
We don't get stuck on minor things. It shouldn't matter if some code can be simplified by a few words or lines if the functionality itself works. It's also fine if the coding style isn't 100% correct. Most important is really that things work. Typos are fine, too.
The goal of submitting and reviewing a PR is to have the code merged, so try to focus on what is left to be done to merge the PR. The only time a review should not give a clear idea of what is left to be done for approval and merging is if the PR has very basic, fundamental issues that prevent it from being properly reviewed (eg the person who submitted it failed to do basic tests and/or the feature doesn't do what it is supposed to do).
Grant approval when any of the following are true:
Of course to anything there can be exceptions. Eg it might not always be worth it to have tests for certain functionality
Remember when replying to team members and community contributors
If we don't merge a PR, produce a record of your decision.
Focus on the work.
When reviewing a pull request in the current milestone, if it works, all review items have been addressed and tests pass, then core developers are allowed to merge it with a squash and merge. Small changes can be merged directly without a review if the developer is 100% certain the change won't have any side effects etc. It is still always recommended to quickly ask another developer that is online to have a look at this PR now as such PRs are quickly reviewed. Once the PR has been merged the branch can be deleted.
If a PR affects the public API in any way a PR should not be merged without a review.
For anything else, we'd have to know whether it is something we want in the current milestone. This "approval" can be an explicit comment from Thomas or Matt, or it could just be from a slack conversation (remember, we can always revert something later). Exceptions can be made, however, if the change is small and not likely to cause any problems when released. Then it's fine to change the milestone to the current one and merge it.
Since we work on issues in the current milestone first, reviewing pull requests outside of the current milestone isn't something that happens very often.