This guide is for people who would like to be involved in building Cats.
This guide assumes that you have some experience doing Scala development. If you get stuck on any of these steps, please feel free to ask for help.
Cats follows a standard fork and pull model for contributions via GitHub pull requests.
Below is a list of the steps that might be involved in an ideal contribution. If you don't have the time to go through every step, contribute what you can, and someone else will probably be happy to follow up with any polishing that may need to be done.
If you want to touch up some documentation or fix typos, feel free to skip these steps and jump straight to submitting a pull request.
- Find something that belongs in cats
- Let us know you are working on it
- Build the project
- Implement your contribution
- Write tests
- Write documentation
- Write examples
- Submit pull request
Looking for a way that you can help out? Check out our open issues and look for ones tagged as help wanted or low-hanging fruit. These issues are the easiest way to start contributing, but if you find other items that catch your eye, you're most than welcome to tackle them!
Make sure that it's not already assigned to someone and that nobody has left a comment saying that they are working on it!
(Of course, you can also comment on an issue someone is already working on and offer to collaborate.)
Have an idea for something new? That's great! We recommend that you make sure it belongs in Cats before you put effort into creating a pull request. The preferred ways to do that are to either:
- create a GitHub issue describing your idea.
- get feedback in the Typelevel Discord
Things that belong in Cats generally have the following characteristics:
- Their behavior is governed by well-defined laws.
- They provide general abstractions.
Laws help keep types consistent, and remove ambiguity or sensitivity about how particular instances can behave. We've found that types with laws are often more useful than lawless types
(In some cases, lawless type classes and instances are useful. We intend to support some of these in a future module.)
By staying general, Cats' abstractions are widely-applicable, and not tied to particular libraries or strategies. Rather than being a library to work with databases, HTTP requests, etc, Cats provides abstractions used to build those libraries.
Cats (and especially cats-core
) is intended to be lean and modular.
Some great ideas are not a great fit, either due to their size or
their complexity. In these cases, creating your own library that
depends on Cats is probably the best plan.
Cats has other companion projects, described next:
- cats-effect: a project aimed to provide a standard IO type for the Cats ecosystem, as well as a set of typeclasses (and associated laws) which characterize general effect types.
- cats-mtl: provides transformer typeclasses for Cats' Monads, Applicatives and Functors.
- mouse: a small companion to the Cats functional programming library for Scala. It includes convenience extension methods for Scala standard library classes, including some found in scalaz that are not in Cats.
If there is already a GitHub issue for the task you are working on, leave a comment to let people know that you are working on it. If there isn't already an issue and it is a non-trivial task, it's a good idea to create one (and note that you're working on it). This prevents contributors from duplicating effort.
First you'll need to checkout a local copy of the code base:
git clone [email protected]:typelevel/cats.git
To build Cats you should have sbt and Node.js installed. If you'd like, you can use the Nix Cats development environment.
Run sbt
, and then use any of the following commands:
compile
: compile the codeconsole
: launch a REPLtest
: run the testsunidoc
: generate the documentationfmt
: run formatting of the codevalidate
: run tests, style-checker, and doc generation
Cats cross-compiles to both JVM and JavaScript runtimes. If you are not used to working with cross-compiling builds, the first things that you will notice is that builds:
-
Will take longer: To build JVM only, just use the
catsJVM
, orcatsJS
for JS only. And if you want the default project to becatsJVM
, just copy the filescripts/sbtrc-JVM
to.sbtrc
in the root directory. -
May run out of memory: We suggest you use Paul Philips's sbt script that will use the settings from Cats.
IntelliJ
-
Be warned, IntelliJ is currently not 100% accurate at reporting compilation errors, there will be cases that it reports errors incorrectly. If you simply don't want to see the errors, a quick an easy work around is to disable Type-Aware Highlighting by clicking the
[T]
icon in the bottom toolbar. -
There is an open issue with the IntelliJ scala plugin, which prevents it from configuring similacrum correctly when importing the cats project. The work around for this issue is to set
val CompileTime = Provided
inbuild.sbt
. Note: Be careful not to commit this change. -
IntelliJ does have support for kind-projector. However, it is not always seamless. If you are unable to get IntelliJ to recognise the special symbols that kind-project provides, such as
?
Lambda[X => G[F[A]]]
orλ[X => G[F[A]]]
try upgrading to the early access preview (EAP) version of the scala plugin. This can be done underSettings > Languages & Frameworks > Scala > Updates
If your contribution has been derived from or inspired by other work, please state this in its ScalaDoc comment and provide proper attribution. When possible, include the original authors' names and a link to the original work.
- Tests for cats-core go into the tests module, under the
cats.tests
package. - Tests for additional modules, such as 'jvm', go into the tests directory within that module.
- Cats tests should extend
CatsSuite
.CatsSuite
integrates with Discipline for law checking, and imports all syntax and standard instances for convenience. - The first parameter to the
checkAll
method provided by Discipline, is the name of the test and will be output to the console as part of the test execution. By convention: - When checking laws, this parameter generally takes a form that describes the data type being tested.
For example the name
"Validated[String, Int]"
might be used when testing a type class instance that theValidated
data type supports. - An exception to this is serializability tests, where the type class name is also included in the name.
For example, in the case of
Validated
, the serializability test would take the form,"Applicative[Validated[String, Int]]"
, to indicate that this test is verifying that theApplicative
type class instance for theValidated
data type is serializable. - This convention helps to ensure clear and easy to understand output, with minimal duplication in the output.
- It is also a goal that, for every combination of data type and supported type class instance:
- Appropriate law checks for that combination are included to ensure that the instance meets the laws for that type class.
- A serializability test for that combination is also included, such that we know that frameworks which
rely heavily on serialization, such as
Spark
, will have strong compatibility withCats
. - Note that custom serialization tests are not required for instances of type classes which come from
algebra
, such asMonoid
, because thealgebra
laws include a test for serialization. - For testing your laws, it is advised to check this guide.
It is important to verify that the feature you are implementing is compatible with Scala 2.12.x and Scala 2.13.x (Scala <2.11.x is not supported). When you submit a PR, Travis makes this check, but it is time-expensive, so you can assure this step beforehand by issuing the command ++2.12.13
, which sets the cats' Scala version to 2.12.13
and then run mimaReportBinaryIssues
.
A summary of these steps is as follows:
$ sbt
> ++2.12.13
> mimaReportBinaryIssues
This procedure will report if there are any binary compatibility issues that should be fixed.
As a side note, the latter command uses sbt-mima (shorthand for "Migration Manager") plugin and you can find more information about it here.
The documentation for this website is stored alongside the source, in the docs subproject.
The command is: sbt docs/tlSite
We suggest checking the CI's workflow to discover any changes at CI site job.
Run docs/tlSitePreview
in the sbt console. This will start a preview server at http://localhost:4242/ that will automatically refresh as you make edits.
We use mdoc to compile source code which appears in the documentation, this ensures us that our examples should always compile, and our documentation has a better chance of staying up-to-date.
One of the best ways to provide examples is doctest, here is an example. Doctest is a sbt plugin which generates tests based on the syntax mentioned above and runs when sbt's test
task is invoked. You can find more information in the plugin documentation.
Before you open a pull request, you should make sure that sbt validate
runs successfully. Travis will run this as well, but it may
save you some time to be alerted to style problems earlier.
If your pull request addresses an existing issue, please tag that issue number in the body of your pull request or commit message. For example, if your pull request addresses issue number 52, please include "fixes #52".
If you make changes after you have opened your pull request, please add them as separate commits and avoid squashing or rebasing. Squashing and rebasing can lead to a tidier git history, but they can also be a hassle if somebody else has done work based on your branch.
Getting involved in an open source project can be tough. As a newcomer, you may not be familiar with coding style conventions, project layout, release cycles, etc. This document seeks to demystify the contribution process for the cats project.
It may take a while to familiarize yourself with this document, but if we are doing our job right, you shouldn't have to spend months poring over the project source code or lurking on Discord before you feel comfortable contributing. In fact, if you encounter any confusion or frustration during the contribution process, please create a GitHub issue and we'll do our best to improve the process.
Discussion around Cats is currently happening in the Typelevel Discord as well as on GitHub issue and PR pages.
Feel free to open an issue if you notice a bug, have an idea for a feature, or have a question about the code. Pull requests are also gladly accepted.
People are expected to follow the Scala Code of Conduct when discussing Cats on GitHub, Discord, or other venues.
We hope that our community will be respectful, helpful, and kind. If you find yourself embroiled in a situation that becomes heated, or that fails to live up to our expectations, you should disengage and contact one of the project maintainers in private. We hope to avoid letting minor aggressions and misunderstandings escalate into larger problems.
If you are being harassed, please contact one of us immediately so that we can support you.
Since Cats development can include the Scala runtime, the Scala.js runtime, the Cats website, and more; a number of dependencies (sbt, Node.js, Jekyll, etc) can be needed to work on Cats. Managing these dependencies globally can be a hassle and can lead to version conflicts. To make this easier to manage in an isolated development environment, Cats provides a shell.nix
for anyone using the Nix package manager.
To use the Nix-based Cats development environment:
- Install the Nix package manager.
- At the root level of the Cats repository, run
nix-shell --pure
. This will drop you into a minimal bash shell that has just the required dependencies on thePATH
. Note that the first time that you run this it will take some extra time to download the necessary dependencies into your local Nix store. - Run
sbt
,jekyll
, etc as required from thenix-shell
. - When you are finished you can
exit
thenix-shell
.