Contributing CodePhabricator Contributor Documentation (Contributing in Detail)
Describes how to contribute code to Phabricator.
To contribute to the Phabricator upstream, you must first pass a series of ancient trials and be invited to register an account in the ancestral homeland of Phabricator, here on secure.phabricator.com. The nature and location of these trials is a closely guarded secret.
If you have passed these trials, this document can guide you through contributing code.
If you have not yet passed these trials, writing code is normally not the best way to contribute to Phabricator. See Contributor Introduction for more information.
If you're planning to send a patch to Phabricator, this guide can help you through the process. The most important parts of contributing code to Phabricator are:
- File a task with a bug report or feature request before you write code.
- We rarely accept patches which we haven't discussed first.
- We do not accept patches against prototype applications.
- You must sign the CLA.
- We do not accept GitHub pull requests.
- Some alternative approaches are available if your change isn't something we want to bring upstream.
The rest of this article describes these points in more detail, and then provides guidance on writing and submitting patches.
If you just want to contribute some code but don't have a specific bug or feature in mind, see the bottom of this document for tips on finding ways to get started.
For general information on contributing to Phabricator, see Contributor Introduction.
Before sending code, you should file a task describing what you'd like to write.
When you file a task, mention that you'd like to write the code to fix it. We can help contextualize your request or bug and guide you through writing an upstreamable patch, provided it's something that's upstreamable. If it isn't upstreamable, we can let you know what the issues are and help find another plan of attack.
You don't have to file first (for example, if you spot a misspelling it's normally fine to just send a diff), but for anything even moderately complex you're strongly encouraged to file first and coordinate with the upstream.
If you send us a patch without coordinating it with us first, it will probably be immediately rejected, or sit in limbo for a long time and eventually be rejected. The reasons we do this vary from patch to patch, but some of the most common reasons are:
Unjustifiable Costs: We support code in the upstream forever. Support is enormously expensive and takes up a huge amount of our time. The cost to support a change over its lifetime is often 10x or 100x or 1000x greater than the cost to write the first version of it. Many uncoordinated patches we receive are "white elephants", which would cost much more to maintain than the value they provide.
As an author, it may look like you're giving us free work and we're rejecting it as too expensive, but this viewpoint doesn't align with the reality of a large project which is actively supported by a small, experienced team. Writing code is cheap; maintaining it is expensive.
By coordinating with us first, you can make sure the patch is something we consider valuable enough to put long-term support resources behind, and that you're building it in a way that we're comfortable taking over.
Not a Good Fit: Many patches aren't good fits for the upstream: they implement features we simply don't want. Coordinating with us first helps make sure we're on the same page and interested in a feature.
The most common type of patch along these lines is a patch which adds new configuration options. We consider additional configuration options to have an exceptionally high lifetime support cost and are very unlikely to accept them. Coordinate with us first.
Not a Priority: If you send us a patch against something which isn't a priority, we probably won't have time to look at it. We don't give special treatment to low-priority issues just because there's code written: we'd still be spending time on something lower-priority when we could be spending it on something higher-priority instead.
If you coordinate with us first, you can make sure your patch is in an area of the codebase that we can prioritize.
Overly Ambitious Patches: Sometimes we'll get huge patches from new contributors. These can have a lot of fundamental problems and require a huge amount of our time to review and correct. If you're interested in contributing, you'll have more success if you start small and learn as you go.
We can help you break a large change into smaller pieces and learn how the codebase works as you proceed through the implementation, but only if you coordinate with us first.
Generality: We often receive several feature requests which ask for similar features, and can come up with a general approach which covers all of the use cases. If you send us a patch for your use case only, the approach may be too specific. When a cleaner and more general approach is available, we usually prefer to pursue it.
By coordinating with us first, we can make you aware of similar use cases and opportunities to generalize an approach. These changes are often small, but can have a big impact on how useful a piece of code is.
Infrastructure and Sequencing: Sometimes patches are written against a piece of infrastructure with major planned changes. We don't want to accept these because they'll make the infrastructure changes more difficult to implement.
Coordinate with us first to make sure a change doesn't need to wait on other pieces of infrastructure. We can help you identify technical blockers and possibly guide you through resolving them if you're interested.
With rare exceptions, we do not accept patches for prototype applications for the same reasons that we don't accept feature requests or bug reports. To learn more about prototype applications, see User Guide: Prototype Applications.
Before we can accept source code contributions, you need to submit a Contributor License Agreement. Your changes can not be accepted until you sign the agreement.
If you haven't signed it by the time you send changes for review, you'll be reminded to sign it at that time.
If you're submitting work on behalf of a company (like your employer), the company can sign the Corporate Contributor License Agreement instead.
Both agreements are substantially similar to the Apache Foundation's CLAs. They protect Phacility and users of Phabricator by making sure we have permission to distribute your changes under an open source license.
We do not accept pull requests on GitHub:
- We can not monitor who has signed CLAs on GitHub. You must sign the CLA to contribute, and we can't tell if you've signed it or not when you send us a pull request.
- Pull requests do not get lint and unit tests run, so issues which are normally caught statically can slip by.
- Phabricator is code review software, and developed using its own workflows. Pull requests bypass some of these workflows (for example, they will not trigger Herald rules to notify interested parties).
- GitHub is not the authoritative master repository and we maintain a linear history, so merging pull requests is cumbersome on our end.
- If you're comfortable enough with Phabricator to contribute to it, you should also be comfortable using it to submit changes.
Instead of sending a pull request, use arc diff to create a revision on the upstream install. Your change will go through the normal Phabricator review process.
(GitHub does not allow repositories to disable pull requests, which is why it's technically possible to submit them.)
If you've written code but we're not accepting it into the upstream, some alternative approaches include:
Maintain a local fork. This will require some ongoing effort to port your changes forward when you update, but is often very reasonable for simple changes.
Develop as an application. Many parts of Phabricator's infrastructure are modular, and modularity is increasing over time. A lot of changes can be built as external modules or applications without forking Phabricator itself. There isn't much documentation or support for this right now, but you can look at how other applications are implemented, and at other third-party code that extends Phabricator.
Rise to prominence. We're more willing to accept borderline changes from community members who are active, make multiple contributions, or have a history with the project. This is not carte blanche, but distinguishing yourself can make us feel more comfortable about supporting a change which is slightly outside of our comfort zone.
To actually submit a patch, run arc diff in phabricator/ or arcanist/. When executed in these directories, arc should automatically talk to the upstream install. You can add epriestley as a reviewer.
You should read the relevant coding convention documents before you submit a change. If you're a new contributor, you don't need to worry about this too much. Just try to make your code look similar to the code around it, and we can help you through the details during review.
- General Coding Standards (for all languages)
- PHP Coding Standards (for PHP)
In general, if you're coordinating with us first, we can usually provide guidance on how to implement things. The other articles in this section also provide information on how to work in the Phabricator codebase.
- returning to the Contributor Introduction.