Welcome! Thank you for contributing to Akka!

We follow the standard GitHub fork & pull approach to pull requests. Just fork the official repo, develop in a branch, and submit a PR!

You're always welcome to submit your PR straight away and start the discussion (without reading the rest of this wonderful doc, or the README.md). The goal of these notes is to make your experience contributing to Akka as smooth and pleasant as possible. We're happy to guide you through the process once you've submitted your PR.

The Akka Community

In case of questions about the contribution process or for discussion of specific issues please visit the akka/dev gitter chat.

You may also check out these other resources.

Navigating around the project & codebase

Branches summary

Depending on which version (or sometimes module) you want to work on, you should target a specific branch as explained below:

master – active development branch of Akka 2.6.x
release-2.5 – maintenance branch of Akka 2.5.x
similarly release-2.# branches contain legacy versions of Akka

Akka contributing guidelines

These guidelines apply to all Akka projects, by which we mean both the akka/akka repository, as well as any plugins or additional repositories located under the Akka GitHub organisation.

These guidelines are meant to be a living document that should be changed and adapted as needed. We encourage changes that make it easier to achieve our goals in an efficient way.

General workflow

The steps below describe how to get a patch into a main development branch (e.g. master). The steps are exactly the same for everyone involved in the project (be it core team, or first time contributor).

To avoid duplicated effort, it might be good to check the issue tracker and existing pull requests for existing work.
- If there is no ticket yet, feel free to create one to discuss the problem and the approach you want to take to solve it.
Fork the project on GitHub. You'll need to create a feature-branch for your work on your fork, as this way you'll be able to submit a pull request against the mainline Akka.
Create a branch on your fork and work on the feature. For example: git checkout -b custom-headers-akka-http
- Please make sure to follow the general quality guidelines (specified below) when developing your patch.
- Please write additional tests covering your feature and adjust existing ones if needed before submitting your pull request. The validatePullRequest sbt task (explained below) may come in handy to verify your changes are correct.
Once your feature is complete, prepare the commit following our Creating Commits And Writing Commit Messages. For example, a good commit message would be: Adding compression support for Manifests #22222 (note the reference to the ticket it aimed to resolve).
If it's a new feature, or a change of behavior, document it on the akka-docs. If the feature was touching Scala or Java DSL, make sure to document both the Scala and Java APIs.
Now it's finally time to submit the pull request!
- Please make sure to include a reference to the issue you're solving in the comment for the Pull Request, as this will cause the PR to be linked properly with the Issue. Examples of good phrases for this are: "Resolves #1234" or "Refs #1234".
If you have not already done so, you will be asked by our CLA bot to sign the Lightbend CLA online. CLA stands for Contributor License Agreement and is a way of protecting intellectual property disputes from harming the project.
A core member will comment OK TO TEST on your PR to kick off the build. This is just a sanity check to prevent malicious code from being run on the Jenkins cluster.
Now both committers and interested people will review your code. This process is to ensure the code we merge is of the best possible quality, and that no silly mistakes slip through. You're expected to follow-up these comments by adding new commits to the same branch. The commit messages of those commits can be more loose, for example: Removed debugging using printline, as they all will be squashed into one commit before merging into the main branch.
- The community and team are really nice people, so don't be afraid to ask follow up questions if you didn't understand some comment, or would like clarification on how to continue with a given feature. We're here to help, so feel free to ask and discuss any kind of questions you might have during review!
After the review you should fix the issues as needed (pushing a new commit for new review etc.), iterating until the reviewers give their thumbs up–which is signalled usually by a comment saying LGTM, which means "Looks Good To Me".
- In general a PR is expected to get 2 LGTMs from the team before it is merged. If the PR is trivial, or under special circumstances (such as most of the team being on vacation, a PR was very thoroughly reviewed/tested and surely is correct) one LGTM may be fine as well.
If the code change needs to be applied to other branches as well (for example a bugfix needing to be backported to a previous version), one of the team will either ask you to submit a PR with the same commit to the old branch, or do this for you.
- Follow the backporting steps below.
Once everything is said and done, your pull request gets merged 🎉 Your feature will be available with the next “earliest” release milestone (i.e. if back-ported so that it will be in release x.y.z, find the relevant milestone for that release). And of course you will be given credit for the fix in the release stats during the release's announcement. You've made it!

The TL;DR; of the above very precise workflow version is:

Fork Akka
Hack and test on your feature (on a branch)
Document it
Submit a PR
Sign the CLA if necessary
Keep polishing it until received enough LGTM
Profit!

Backporting

Backport pull requests such as these are marked using the phrase for validation in the title to make the purpose clear in the pull request list. They can be merged once validation passes without additional review (if there are no conflicts). Using, for example: current.version 2.5.22, previous.version 2.5, milestone.version 2.6.0-M1

Label this PR with to-be-backported
Mark this PR with Milestone ${milestone.version}
Mark the issue with Milestone ${current.version}
git checkout release-${previous.version}
git pull
Create wip branch
git cherry-pick <commit>
Open PR, target release-${previous.version}
Label that PR with backport
Merge backport PR after validation (no need for full PR reviews)
Close issue

Getting started with sbt

Akka is using the sbt build system. So the first thing you have to do is to download and install sbt. You can read more about how to do that in the sbt setup documentation.

To compile all the Akka core modules use the compile command:

sbt compile

You can run all tests with the test command:

sbt test

If you want to deploy the artifacts to your local Ivy repository (for example, to use from an sbt project) use the publishLocal command:

sbt publishLocal

Note that in the examples above we are calling sbt compile and sbt test and so on, but sbt also has an interactive mode. If you just run sbt you enter the interactive sbt prompt and can enter the commands directly. This saves starting up a new JVM instance for each command and can be much faster and more convenient.

For example, building Akka as above is more commonly done like this:

% sbt
[info] Set current project to default (in build file:/.../akka/project/plugins/)
[info] Set current project to akka (in build file:/.../akka/)
> compile
...
> test
...

To run a single multi-jvm test:

sbt
project akka-cluster
multi-jvm:testOnly akka.cluster.SunnyWeather

To format the Scala source code:

sbt
akka-cluster/scalafmtAll
akka-persistence/scalafmtAll

Do not use `-optimize` Scala compiler flag

Akka has not been compiled or tested with -optimize Scala compiler flag. (In sbt, you can specify compiler options in the scalacOptions key.) Strange behavior has been reported by users that have tried it.

Compiling with Graal JIT

Akka, like most Scala projects, compiles faster with the Graal JIT enabled. The easiest way to use it for compiling Akka is to:

Use a JDK > 10
Use the following JVM options for SBT e.g. by adding them to the SBT_OPTS environment variable: -XX:+UnlockExperimentalVMOptions -XX:+EnableJVMCI -XX:+UseJVMCICompiler

The `validatePullRequest` task

The Akka build includes a special task called validatePullRequest which investigates the changes made as well as dirty (uncommitted changes) in your local working directory and figures out which projects are impacted by those changes, then running tests only on those projects.

For example changing something in akka-actor would cause tests to be run in all projects which depend on it (e.g. akka-actor-tests, akka-stream, akka-docs etc.).

To use the task simply type validatePullRequest, and the output should include entries like shown below:

> validatePullRequest
[info] Diffing [HEAD] to determine changed modules in PR...
[info] Detected uncomitted changes in directories (including in dependency analysis): [akka-protobuf,project]
[info] Detected changes in directories: [akka-actor-tests, project, akka-stream, akka-docs, akka-persistence]

By default changes are diffed with the master branch when working locally, if you want to validate against a different target PR branch you can do so by setting the PR_TARGET_BRANCH environment variable for sbt:

PR_TARGET_BRANCH=origin/example sbt validatePullRequest

If you have already run all tests and now just need to check that everything is formatted and or mima passes there are a set of all* commands aliases for running test:compile (also formats), mimaReportBinaryIssues, and validateCompile (compiles multi-jvm if enabled for that project). See build.sbt or use completion to find the most appropriate one e.g. allCluster, allTyped.

Binary compatibility

Binary compatibility rules and guarantees are described in depth in the Binary Compatibility Rules section of the documentation.

Akka uses MiMa to validate binary compatibility of incoming pull requests. If your PR fails due to binary compatibility issues, you may see an error like this:

[info] akka-stream: found 1 potential binary incompatibilities while checking against com.typesafe.akka:akka-stream_2.12:2.4.2  (filtered 222)
[error]  * method foldAsync(java.lang.Object,scala.Function2)akka.stream.scaladsl.FlowOps in trait akka.stream.scaladsl.FlowOps is present only in current version
[error]    filter with: ProblemFilters.exclude[ReversedMissingMethodProblem]("akka.stream.scaladsl.FlowOps.foldAsync")

In such situations it's good to consult with a core team member whether the violation can be safely ignored or if it would indeed break binary compatibility. If the violation can be ignored add exclude statements from the mima output to a new file named <module>/src/main/mima-filters/<last-version>.backwards.excludes/<pr-or-issue>-<issue-number>-<description>.excludes, e.g. akka-actor/src/main/mima-filters/2.6.0.backwards.excludes/pr-12345-rename-internal-classes.excludes. Make sure to add a comment in the file that describes briefly why the incompatibility can be ignored.

Situations when it may be fine to ignore a MiMa issued warning include:

if it is touching any class marked as private[akka], /** INTERNAL API*/ or similar markers
if it is concerning internal classes (often recognisable by package names like dungeon, impl, internal etc.)
if it is adding API to classes / traits which are only meant for extension by Akka itself, i.e. should not be extended by end-users
other tricky situations

The binary compatibility of the current changes can be checked by running sbt +mimaReportBinaryIssues.

Wire compatibility

Changes to the binary protocol of remoting, cluster and the cluster tools require great care so that it is possible to do rolling upgrades. Note that this may include older nodes communicating with a newer node so compatibility may have to be both ways.

Since during a rolling upgrade nodes producing the 'new' format and nodes producing the 'old' format coexist, a change can require a two-release process: the first change is to add a new binary format but still use the old. A second step then starts actually emitting the new wire format. This ensures users can complete a rolling upgrade first to the intermediate version and then another rolling upgrade to the next version.

All wire protocol changes that may concern rolling upgrades should be documented in the Rolling Update Changelog (found in akka-docs/src/main/paradox/project/rolling-update.md)

Protobuf

Akka includes a shaded version of protobuf 3.9.0 that is used for internal communication. To generate files run protobufGenerate. The generated files are put in the src/main/java of each project and need to be committed. The generated files are automatically transformed to use the shaded version of protobuf.

Generation depends on protoc 3.9.0 being on the path. Old versions of protoc can be downloaded from the protobuf release page and built from source or downloaded from maven central. See Protobuf.scala for details of how to override the settings for generation.

Pull request requirements

For a pull request to be considered at all it has to meet these requirements:

Regardless if the code introduces new features or fixes bugs or regressions, it must have comprehensive tests.
The code must be well documented in the Lightbend's standard documentation format (see the ‘Documentation’ section below).
The commit messages must properly describe the changes, see further below.
A pull request must indicate (link to) the issue it is aimed to resolve in the description (or comments) of the PR, in order to establish a link between PR and Issue. This can be achieved by writing "Fixes #1234" or similar in PR description.
All Lightbend projects must include Lightbend copyright notices. Each project can choose between one of two approaches:
1. All source files in the project must have a Lightbend copyright notice in the file header.
2. The Notices file for the project includes the Lightbend copyright notice and no other files contain copyright notices. See http://www.apache.org/legal/src-headers.html for instructions for managing this approach for copyrights.
Akka uses the first choice, having copyright notices in every file header. When absent, these are added automatically during sbt compile.

Additional guidelines

Some additional guidelines regarding source code are:

Keep the code DRY.
Apply the Boy Scout Rule whenever you have the chance to.
Never delete or change existing copyright notices, just add additional info.
Do not use @author tags since it does not encourage Collective Code Ownership.
- Contributors , each project should make sure that the contributors gets the credit they deserve—in a text file or page on the project website and in the release notes etc.

Documentation

All documentation is preferred to be in Lightbend's standard documentation format Paradox, which among other things allows all code in the documentation to be externalized into compiled files and imported into the documentation.

To build the documentation locally:

sbt
akka-docs/paradox

The generated html documentation is in akka-docs/target/paradox/site/main/index.html.

Alternatively, use akka-docs/paradoxBrowse to open the generated docs in your default web browser.

Links to API documentation

Akka Paradox supports directives to link to the Scaladoc- and Javadoc-generated API documentation:

@apidoc[Flow] searches for the class name and creates links to Scaladoc and Javadoc (see variants in sbt-paradox-apidoc)
@scaladoc[Flow](akka.stream.scaladsl.Flow) (see Paradox docs)
@javadoc[Flow](akka.stream.javadsl.Flow) (see Paradox docs)

Scaladoc

Akka generates class diagrams for the API documentation using ScalaDoc.

Links to methods in ScalaDoc comments should be formatted [[Like#this]], because [[this]] does not work with genjavadoc, and IntelliJ warns about [[#this]]. For further hints on how to disambiguate links in scaladoc comments see this StackOverflow answer, though note that this syntax may not correctly render as Javadoc.

The Scaladoc tool needs the dot command from the Graphviz software package to be installed to avoid errors. You can disable the diagram generation by adding the flag -Dakka.scaladoc.diagrams=false. After installing Graphviz, make sure you add the toolset to the PATH (definitely on Windows).

JavaDoc

Akka generates JavaDoc-style API documentation using the genjavadoc sbt plugin, since the sources are written mostly in Scala.

Generating JavaDoc is not enabled by default, as it's not needed on day-to-day development as it's expected to just work. If you'd like to check if your links and formatting look good in JavaDoc (and not only in ScalaDoc), you can generate it by running:

sbt -Dakka.genjavadoc.enabled=true javaunidoc:doc

Which will generate JavaDoc style docs in ./target/javaunidoc/index.html. This requires a jdk version 9 or later.

External dependencies

All the external runtime dependencies for the project, including transitive dependencies, must have an open source license that is equal to, or compatible with, Apache 2.

This must be ensured by manually verifying the license for all the dependencies for the project:

Whenever a committer to the project changes a version of a dependency (including Scala) in the build file.
Whenever a committer to the project adds a new dependency.
Whenever a new release is cut (public or private for a customer).

Which licenses are compatible with Apache 2 are defined in this doc, where you can see that the licenses that are listed under Category A are automatically compatible with Apache 2, while the ones listed under Category B need additional action:

Each license in this category requires some degree of reciprocity; therefore, additional action must be taken in order to minimize the chance that a user of an Apache product will create a derivative work of a reciprocally-licensed portion of an Apache product without being aware of the applicable requirements.

Each project must also create and maintain a list of all dependencies and their licenses, including all their transitive dependencies. This can be done either in the documentation or in the build file next to each dependency.

Creating commits and writing commit messages

Follow these guidelines when creating public commits and writing commit messages.

If your work spans multiple local commits (for example; if you do safe point commits while working in a feature branch or work in a branch for a long time doing merges/rebases etc.) then please do not commit it all but rewrite the history by squashing the commits into a single big commit which you write a good commit message for (like discussed in the following sections). For more info read this article: Git Workflow. Every commit should be able to be used in isolation, cherry picked etc.
The first line should be a descriptive sentence what the commit is doing, including the ticket number. It should be possible to fully understand what the commit does—but not necessarily how it does it—by just reading this single line. We follow the “imperative present tense” style for commit messages (more info here).

It is not ok to only list the ticket number, type "minor fix" or similar. If the commit is a small fix, then you are done. If not, go to 3.
Following the single line description should be a blank line followed by an enumerated list with the details of the commit.
You can request review by a specific team member for your commit (depending on the degree of automation we reach, the list may change over time):
- Review by @gituser - if you want to notify someone on the team. The others can, and are encouraged to participate.

Example:

enable Travis CI #1

* Details 1
* Details 2
* Details 3

Pull request validation workflow details

Akka uses Jenkins GitHub pull request builder plugin that automatically merges the code, builds it, runs the tests and comments on the pull request in GitHub.

Upon a submission of a pull request the GitHub pull request builder plugin will post a following comment:

Can one of the repo owners verify this patch?

This requires a member from a core team to start the pull request validation process by posting a comment consisting only of OK TO TEST. From now on, whenever new commits are pushed to the pull request, a validation job will be automatically started and the results of the validation posted to the pull request.

A pull request validation job can be started manually by posting PLS BUILD comment on the pull request.

In order to speed up PR validation times, the Akka build contains a special sbt task called validatePullRequest, which is smart enough to figure out which projects should be built if a PR only has changes in some parts of the project. For example, if your PR only touches akka-persistence, no akka-remote tests need to be run, however the task will validate all projects that depend on akka-persistence (including samples). Also, tests tagged as PerformanceTest, TimingTest, LongRunningTest and all multi-node tests are excluded from PR validation.

You can exclude the same kind of tests in your local build by starting sbt with:

sbt -Dakka.test.tags.exclude=performance,timing,long-running -Dakka.test.multi-in-test=false

In order to force the validatePullRequest task to build the entire project, regardless of dependency analysis of a PRs changes one can use the special PLS BUILD ALL command (typed in a comment on GitHub, on the pull request), which will cause the validator to test all projects.

Note, that OK TO TEST will only be picked up when the user asking for it is considered an admin. Public (!) members of the akka organization are automatically considered admins and users manually declared admin in the Jenkins job (currently no one is explicitly listed). PLS BUILD and PLS BUILD ALL can be issued by everyone that is an admin or everyone who was whitelisted in the Jenkins Job (whitelisting != declaring someone an admin).

Source style

Sometimes it is convenient to place 'internal' classes in their own package. In such situations we prefer 'internal' over 'impl' as a package name.

Scala style

Akka uses Scalafmt to enforce some of the code style rules.

It's recommended to enable Scalafmt formatting in IntelliJ. Use version 2019.1 or later. In Preferences > Editor > Code Style > Scala, select Scalafmt as formatter and enable "Reformat on file save". IntelliJ will then use the same settings and version as defined in .scalafmt.conf file. Then it's not needed to use sbt scalafmtAll when editing with IntelliJ.

PR validation includes checking that the Scala sources are formatted and will fail if they are not.

Java style

Akka uses the sbt Java Formatter plugin to format Java sources.

PR validation includes checking that the Java sources are formatted and will fail if they are not.

Code discipline opt out

In addition to formatting the Akka build enforces code discipline through a set of compiler flags. While exploring ideas the discipline may be more of a hindrance than a help so it is possible to disable it by setting the system property akka.no.discipline to any non-empty string value when starting up sbt: sbt -Dakka.no.discipline=youbet

PR validation includes the discipline flags and therefore may fail if the flags were disabled during development. Make sure you compile your code at least once with discipline enabled before sending a PR.

Preferred ways to use timeouts in tests

Avoid short test timeouts, since Jenkins server may GC heavily causing spurious test failures. GC pause or other hiccups of 2 seconds are common in our CI environment. Please note that usually giving a larger timeout does not slow down the tests, as in an expectMessage call for example it usually will complete quickly.

There are a number of ways timeouts can be defined in Akka tests. The following ways to use timeouts are recommended (in order of preference):

remaining is first choice (requires within block)
remainingOrDefault is second choice
3.seconds is third choice if not using testkit
lower timeouts must come with a very good reason (e.g. Awaiting on a known to be "already completed" Future)

Special care should be given to expectNoMessage calls, which indeed will wait the entire timeout before continuing, therefore a shorter timeout should be used in those, for example 200 or 300.millis. Prefer the method without timeout parameter, which will use the configured expect-no-message-default timeout.

You can read up on remaining and friends in TestKit.scala.

Contributing modules

For external contributions of entire features, the normal way is to establish it as a stand-alone feature first, to show that there is a need for the feature. The next step would be to add it to Akka as an "may change"-feature (in the akka-contrib subproject) and marking it's public api with the ApiMayChange annotation, then when the feature is hardened, well documented and tested it becomes an officially supported Akka feature.

List of Akka features marked as may change

Java APIs in Akka

Akka, aims to keep 100% feature parity between the Java and Scala. Implementing even the API for Java in Scala has proven the most viable way to do it, as long as you keep the following in mind:

Keep entry points separated in javadsl and scaladsl unless changing existing APIs which for historical and binary compatibility reasons do not have this subdivision.
Have methods in the javadsl package delegate to the methods in the Scala API, or the common internal implementation. The Akka Stream Scala instances for example have a .asJava method to convert to the akka.stream.javadsl counterparts.
When using Scala object instances, offer a getInstance() method. See akka.Done for an example.
When the Scala API contains an apply method, use create or of for Java users.
Do not nest Scala objects more than two levels.
Do not define traits nested in other classes or in objects deeper than one level.
Be careful to convert values within data structures (eg. for scala.Long vs. java.lang.Long, use scala.Long.box(value))
Complement any methods with Scala collections with a Java collection version
Use the akka.japi.Pair class to return tuples
If the underlying Scala code requires an ExecutionContext, make the Java API take an Executor and use ExecutionContext.fromExecutor(executor) for conversion.
Make use of scala-java8-compat conversions, see GitHub (eg. scala.compat.java8.FutureConverters to translate Futures to CompletionStages). Note that we cannot upgrade to a newer version scala-java8-compat because of binary compatibility issues.
Make sure there are Java tests or sample code touching all parts of the API
Do not use lower type bounds: trait[T] { def method[U >: Something]: U } as they do not work with Java
Provide getX style accessors for values in the Java APIs
Place classes not part of the public APIs in a shared internal package. This package can contain implementations of both Java and Scala APIs. Make such classes private[akka] and also, since that becomes public from Java's point of view, annotate with @InternalApi and add a scaladoc saying INTERNAL API
Traits that are part of the Java API should only be used to define pure interfaces, as soon as there are implementations of methods, prefer abstract class.
Any method definition in a class that will be part of the Java API should not use any default parameters, as they will show up ugly when using them from Java, use plain old method overloading instead.

Overview of Scala types and their Java counterparts

Scala	Java
`scala.Option[T]`	`java.util.Optional<T>` (`OptionalDouble`, ...)
`scala.collection.immutable.Seq[T]`	`java.util.List<T>`
`scala.concurrent.Future[T]`	`java.util.concurrent.CompletionStage<T>`
`scala.concurrent.Promise[T]`	`java.util.concurrent.CompletableFuture<T>`
`scala.concurrent.duration.FiniteDuration`	`java.time.Duration` (use `akka.util.JavaDurationConverters`)
`T => Unit`	`java.util.function.Consumer<T>`
`() => R` (`scala.Function0[R]`)	`java.util.function.Supplier<R>`
`T => R` (`scala.Function1[T, R]`)	`java.util.function.Function<T, R>`

Contributing new Akka Streams operators

Documentation of Akka Streams operators is automatically enforced. If a method exists on Source / Sink / Flow, or any other class listed in project/StreamOperatorsIndexGenerator.scala, it must also have a corresponding documentation page under akka-docs/src/main/paradox/streams/operators/....

The pages structure is well-defined, and must be the same on all documentation pages, please refer to any neighbouring docs pages in there to see the pattern in action. In general though the page must consist of:

the title, including where the operator is defined (e.g. ActorFlow.ask or Source.map)
a short explanation of what this operator does, 1 sentence is optimal
an image explaining the operator more visually (whenever possible)
a link to the operators' "category" (these are listed in akka-docs/src/main/paradox/categories)
the method signature snippet (use the built in directives to generate it)
a longer explanation about the operator and it's exact semantics (when it pulls, cancels, signals elements)
at least one usage example

Using this structure, the surrounding infrastructure will generate the index pages, so you do not need to maintain the index or category pages manually.

Adding new top-level objects/classes containing operators

In case you are adding not only a new operator, but also a new class/object, you need to add it to the project/StreamOperatorsIndexGenerator.scala so it can be included in the automatic docs generation and enforcing the existence of those docs.

Supporting infrastructure

Reporting security issues

If you have found an issue in an Akka project that might have security implications, you can report it to security@lightbend.com. We will make sure those will get handled with priority. Thank you for your responsible disclosure!

Continuous integration

Akka currently uses a combination of Jenkins and Travis for Continuous Integration:

Jenkins runs the tests for each PR
Jenkins runs a nightly test suite
Travis checks dependency licenses for all PR's

The Jenkins server farm, sometimes referred to as "the Lausanne cluster", is sponsored by Lightbend.

The cluster is made out of real bare-metal boxes, and maintained by the Akka team (and other very helpful people at Lightbend).

Files

CONTRIBUTING.md

Latest commit

History