Thank you for your interest in contributing to FAKE! This guide explains everything you'll need to know to get started.
Before diving in, please note:
You are encouraged to improve this document by sending a pull request to the FAKE project on GitHub. If you learn something while playing with FAKE, please record your findings here!
If you'd like to discuss a feature (a good idea!) or are looking for suggestions on how to to contribute:
Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the Project shall be under the terms and conditions of the Apache 2.0 license. See
Before building and developing FAKE, you must:
Linux and Mac users should install the .NET Core SDK and Mono per this guide, "Cross-Platform Development with F#".
Windows users can install Visual Studio 2017. The Community Edition is freely available for open-source projects.
INFOWhen developing on Windows, make sure to have long paths enabled (instructions available here), otherwise the test-suite will fail -- although, the build should work.
You can quickly install FAKE using the dotnet SDK:
dotnet tool install fake-cli -g
For alternative methods of installing FAKE, please see the Getting Started guide.
Fork the FAKE repo on GitHub.
Clone your personal fork locally.
Add a new git remote in order to retrieve upstream changes.
git remote add upstream https://github.com/fsharp/FAKE.git
git checkout master
To verify that everything works, build
fake run build.fsx
Create a new feature branch.
git checkout -b myfeature
Implement your bugfix/feature.
Add a bit of documentation (see the section on Contributing Documentation).
Re-run the build script to confirm that all tests pass.
fake run build.fsx
Commit your changes, and push them to your fork.
Use GitHub's UI to create a pull request. (Write "WIP" into the pull request description if it's not completely ready.)
If you need to rebase you can do:
git fetch upstream git rebase upstream/master git push origin myfeature -f
The pull request will be updated automatically.
The code for all documentation can be found in the
help directory on GitHub. If you find a bug or add a new feature, make sure you document it!
Documentation for FAKE is automatically generated using the amazing F# Formatting library.
It turns Markdown files
*.md with embedded code snippets and F# script
*.fsx files containing embedded Markdown documentation into nice HTML documentation.
To build the documentation from scratch, simply run:
fake build target GenerateDocs
To save time, you may skip the prerequisite build steps and run the
GenerateDocs target directly using the single target
(Note: this assumes binaries are already built and have not been modified.)
fake build -s target GenerateDocs
Running the following target spins up a webserver on localhost and opens the newly built docs in a browser window:
fake build target HostDocs
If you make a change to a module and would like to test it in a fake script, the easiest way to do this is to create a local nuget package and reference it in your script per the steps below:
Create a local nuget package for the module you've changed.
e.g: Using dotnet cli
cd path/to/project dotnet pack
dotnet packwill create a default nuget package with version of 1.0.0 in the
bin/Debugof your project. Set an additional paket source in your build script to this directory, and require this exact version in your paket references.
e.g: If you wanted to test a local build of Fake.DotNet.NuGet
1: 2: 3: 4: 5:
#r "paket: source path/to/Fake.DotNet.NuGet/bin/Debug/ source https://api.nuget.org/v3/index.json ...Other Dependencies... nuget Fake.DotNet.NuGet == 1.0.0 //" //Require version 1.0.0, which is the local build
Read the F# component design guidelines.
Read the API Design Guidelines below.
- Add documentation for your feature
If you add new markdown documentation, make sure to link if from an existing site. Ideally, add it to the menu
- If you write API documentation but no extra markdown, please consider adding it to the menu as well.
We've learned from our mistakes and adopted new API design guidelines. Please read them very carefully, and please ask if you don't understand any of the following rules:
[<AutoOpen>]is no longer used
- Use Verbs as much as possible for functions
- In order to have a more consistent API, we propose to always use camelCase naming for functions
- For historic reasons, we allow constants and values in PascalCase. (They will not have a "verb" as they don't do anything)
If we provide optional parameters (via
static member), we use PascalCase as well (Example:
We assume the caller is not opening the module but only the global namespaces
Fake.IO, ... and make sure the code looks nice and structured on the caller side.
For compatibility reasons (migration from legacy), we assume the user doesn't open the global
-> This means we don't add anything in there in the new API.
Old APIs are marked as Obsolete with a link (hint) to the new API location. We use codes to make explicit
- FAKE0001 for moving part from one Module to another
- FAKE0002 for removed API we don't know who is using it and how => please open an issue if you use it
- FAKE0003 for API that is no more accessible (basically became internal) => please open an issue if you use it
- FAKE0004 for API not yet migrated, waiting for your contribution
Operators are opened seperatly with a separate
We avoid the
Helperssuffix (because we now expect users to write
We generally use the
[<RequireQualifiedAccess>]attribute on modules.
- Fake 4 (FakeLib) is in maintainance mode. Therefore new features need to be at least available as new FAKE 5 module (that might mean that the old module needs to be migrated as part of the PR).
Fake 4 still allows hotfixes. Please send the PR against the hotfix_fake4 branch.
It would be helpful if a second PR against
masteris sent that merges the hotfix into
masterand adds the hotfix to the FAKE 5 code as well.
As mentioned in the "Fake 5 learn more" section, we could use your help porting modules to FAKE 5. To save you from some pitfalls, this section provides a working approach to migrating modules.
Tooling in netcore is not optimal yet, so some things have to be done by hand. These steps will enable pretty good IDE support:
- Copy one of the existing netcore projects and edit the project file by hand (rename)
- Copy the old implementation files from
/src/app/Fake.<ModuleType>.<Name>(update project file again if required)
- Reference the new files in FakeLib (again updating
FakeLib.fsprojby hand to properly reference the stuff)
Fake.slnand go from there. Because in F# you can only reference stuff defines in files ABOVE, this is ALMOST perfect
- Once stuff compiles in the (
Fake.sln) solution the remaining changes to make the netcore project compile are usually straightforward (you basically only need to fix project references or add framework nuget packages if needed). Let us know if you struggle at this point (in the PR or a new issue).
- Add the info about the new module to the
build.fsx. From this point on the build script will let you know if anything is missing. Again, if you have problems let us know.
Mark the old module with the
src/Fake-netcore.slnis currently not used (as IDEs don't support that yet). However it is used so speed up the build,
fake run build.fsxwill let you know what to do in the error message.
These steps will ensure:
- People using the NuGet package will get the warnings to update the new API
- The new API is part of FakeLib (deprecated)
- The new API is available as separate module
In order to test and preview our changes faster, we have a fully automated release process in place. This staging environment is based on VSTS and MyGet.
If you ever need a release/bugfix, make sure to mention that in your PR. We can quickly provide a build on the following infrastructure:
- Website: https://staging.fake.build
- Chocolatey package:
choco install fake --version <version> --source https://www.myget.org/F/fake-chocolatey-vsts/api/v2
- NuGet feed: https://www.myget.org/F/fake-vsts/api/v3/index.json
- GitHub Releases: https://github.com/fake-staging/FAKE/releases (if needed)
INFOBecause of package retention policies those builds will not be available forever! We will quickly release the builds once everything works. Those bits should be considered for "unblocking"-purposes or testing only.
The release process is publicly available as well.