lihaoyi / mill

Your shiny new Java/Scala build tool!

Website GitHub

Mill Logo Mill Build Status Build (Windows) Gitter Chat Patreon

Your shiny new Scala build tool! Confused by SBT? Frustrated by Maven? Perplexed by Gradle? Give Mill a try!

If you want to use Mill in your own projects, check out our documentation:

The remainder of this readme is developer-documentation targeted at people who wish to work on Mill's own codebase. The developer docs assume you have read through the user-facing documentation linked above. It's also worth spending a few minutes reading the following blog posts to get a sense of Mill's design & motivation:

How to build and test

Mill is built using Mill. To begin, first download & install Mill as described in the documentation above. As Mill is under active development, stable releases may not be able to build the current development branch of Mill. It is recommended to install the latest unstable release manually.

IntelliJ Setup

If you are using IntelliJ IDEA to edit Mill's Scala code, you can create the IntelliJ project files via:

mill mill.scalalib.GenIdea/idea

Automated Tests

To run test suites:

mill main.test
mill scalalib.test
mill scalajslib.test
mill integration.test

Manual Testing

To manually test Mill on a small build, you can use the scratch folder:

mill -i scratch -w show thingy

This runs your current checkout of Mill on the trivial build defined in scratch/ You can modify that build file to add additional modules, files, etc. and see how it behaves.

More generally, you can use:

mill -i [target-dir] [...args]

To create run your current checkout of Mill in the given target-dir with the given args. This is useful e.g. to test a modified version of Mill on some other project's Mill build.

You can also create a launcher-script to let you run the current checkout of Mill without the bootstrap Mill process present:

mill -i dev.launcher

This creates the out/dev/launcher/dest/run launcher script, which you can then use to run your current checkout of Mill where-ever you'd like. Note that this script relies on the compiled code already present in the Mill out/ folder, and thus isn't suitable for testing on Mill's own Mill build since you would be over-writing the compiled code at the same time as the launcher script is using it.

You can also run your current checkout of Mill on the build in your scratch/ folder without the bootstrap Mill process being present via:

mill -i dev.launcher && (cd scratch && ../out/dev/launcher/dest/run -w show thingy)

Bootstrapping: Building Mill with your current checkout of Mill

To test bootstrapping of Mill's own Mill build using a version of Mill built from your checkout, you can run


This creates a standalone assembly at ~/mill-release you can use, which references jars published locally in your ~/.ivy2/local cache. You can then use this standalone assembly to build & re-build your current Mill checkout without worrying about stomping over compiled code that the assembly is using.


In case of troubles with caching and/or incremental compilation, you can always restart from scratch removing the out directory:

rm -rf out/

Project Layout

The Mill project is organized roughly as follows:

Core modules that are included in the main assembly

  • core, main, main.client, scalalib, scalajslib.

These are general lightweight and dependency-free: mostly configuration & wiring of a Mill build and without the heavy lifting.

Heavy lifting is delegated to the worker modules (described below), which the core modules resolve from Maven Central (or from the local filesystem in dev) and load into isolated classloaders.

Worker modules that are resolved from Maven Central

  • scalalib.worker, scalajslib.jsbridges[0.6], scalajslib.jsbridges[1.0]

These modules are where the heavy-lifting happens, and include heavy dependencies like the Scala compiler, Scala.js optimizer, etc.. Rather than being bundled in the main assembly & classpath, these are resolved separately from Maven Central (or from the local filesystem in dev) and kept in isolated classloaders.

This allows a single Mill build to use multiple versions of e.g. the Scala.js optimizer without classpath conflicts.



  • Added the mill show visualize command, making it easy to visualize the relationships between various tasks and modules in your Mill build.

  • Improve Intellij support (351): better jump-to-definition for third-party libraries, no longer stomping over manual configuration, and better handling of import $ivy in your build file.

  • Support for un-signed publishing and cases where your GPG key has no passphrase (346)

  • Basic support for Twirl, Play Framework's templating language (271)

  • Better performance for streaming large amounts of stdout from Mill's daemon process.

  • Allow configuration of append/exclude rules in ScalaModule#assembly (309)


  • Preserve caches when transitioning between -i/--interactive and the fast client/server mode (329)

  • Keep Mill daemon running if you Ctrl-C during -w/--watch mode (327)

  • Allow mill version to run without a build file (328)

  • Make docJar (and thus publishing) robust against scratch files in the source directories (334) and work with Scala compiler options (336)

  • Allow passing Ammonite command-line options to the foo.repl command (333)

  • Add mill clean (315) to easily delete the Mill build caches for specific targets

  • Improve IntelliJ integration of MavenModules/SbtModules' test folders (298)

  • Avoid showing useless stack traces when foo.test result-reporting fails or fails

  • ScalaFmt support (308)

  • Allow ScalaModule#generatedSources to allow single files (previous you could only pass in directories)


  • Universal (combined batch/sh) script generation for launcher, assembly, and release (#264)

  • Windows client/server improvements (#262)

  • Windows repl support (note: MSYS2 subsystem/shell will be supported when jline3 v3.6.3 is released)

  • Fixed Java 9 support

  • Remove need for running publishAll using --interactive when on OSX and your GPG key has a passphrase

  • First-class support for JavaModules

  • Properly pass compiler plugins to Scaladoc (#282)

  • Support for ivy version-pinning via ivy"...".forceVersion()

  • Support for ivy excludes via ivy"...".exclude() (#254)

  • Make ivyDepsTree properly handle transitive dependencies (#226)

  • Fix handling of runtime-scoped ivy dependencies (#173)

  • Make environment variables available to Mill builds (#257)

  • Support ScalaCheck test runner (#286)

  • Support for using Typelevel Scala (#275)

  • If a module depends on multiple submodules with different versions of an ivy dependency, only one version is resolved (#273)


  • Support for non-interactive (client/server) mode on Windows.

  • More fixes for Java 9

  • Bumped the Mill daemon timeout from 1 minute to 5 minutes of inactivity before it shuts down.

  • Avoid leaking Node.js subprocesses when running ScalaJSModule tests

  • Passing command-line arguments with spaces in them to tests no longer parses wrongly

  • ScalaModule#repositories, scalacPluginIvyDeps, scalacOptions, javacOptions are now automatically propagated to Tests modules

  • ScalaJSModule linking errors no longer show a useless stack trace

  • ScalaModule#docJar now properly uses the compileClasspath rather than runClasspath

  • Bumped underlying Ammonite version to 1.1.0, which provides the improved Windows and Java 9 support


  • Fixes for non-interactive (client/server) mode on Java 9

  • Windows batch (.bat) generation for launcher, assembly, and release


  • Introduced the mill plan command, which shows you what the execution plan of running the task looks like without actually evaluating it.

  • Mill now generates an out/mill-profile.json file containing task-timings, to make it easier to see where your mill evaluation time is going

  • Introduced ScalaModule#ivyDepsTree command to show dependencies tree

  • Rename describe to inspect for consistency with SBT

  • mill resolve now prints results sorted alphabetically

  • Node.js configuration can be customised with ScalaJSModule#nodeJSConfig

  • Scala.js fullOpt now uses Google Closure Compiler after generating the optimized Javascript output

  • Scala.js now supports NoModule and CommonJSModule module kinds

  • Include compileIvyDeps when generating IntelliJ projects

  • Fixed invalid POM generation

  • Support for Java 9 (and 10)

  • Fixes for Windows support

  • Fixed test classes discovery by skipping interfaces

  • Include "optional" artifacts in dependency resolution if they exist

  • out/{module_name} now added as a content root in generated IntelliJ project


  • Speed up Mill client initialization by another 50-100ms

  • Speed up incremental assemblys in the common case where upstream dependencies do not change.

  • Make ScalaJSModule#run work with main-method discovery

  • Make ScalaWorkerModule user-defineable, so you can use your own custom coursier resolvers when resolving Mill's own jars

  • Simplify definitions of SCM strings

  • Make the build REPL explicitly require -i/--interactive to run

  • Log a message when Mill is initializing the Zinc compiler interface


  • Greatly reduced the overhead of evaluating Mill tasks, with a warm already-cached mill dev.launcher now taking ~450ms instead of ~1000ms

  • Mill now saves compiled build files in ~/.mill/ammonite, which is configurable via the --home CLI arg.

  • Fixed linking of multi-module Scala.js projects


  • Mill now keeps a long-lived work-daemon around in between commands; this should improve performance of things like compile which benefit from the warm JVM. You can use -i/--interactive for interactive consoles/REPLs and for running commands without the daemon

  • Implemented the ScalaModule#launcher target for easily creating command-line launchers you can run outside of Mill

  • ScalaModule#docJar no longer fails if you don't have scala-compiler on classpath

  • Support for multiple testFrameworks in a test module.


  • Fixes for foo.console
  • Enable Ammonite REPL integration via foo.repl


  • First public release