Installation

No published package yet

Invaract is early-stage and hasn’t published a Maven Central artifact yet. If you’re contributing to Invaract itself, or want to build the example harness and web UI, you use it by cloning the repository and building its modules from source — either on your own machine or, if you’d rather not install the toolchain yourself, in GitHub Codespaces. Both paths are covered below.

If you just want to install the verification engine into a Spark job of your own and your build tool isn’t sbt (Maven, Gradle, …), see Use Prebuilt Jars Without sbt instead — it skips building from source entirely.

Prerequisites

Tool	Version	Needed for
JDK	21	Building and running every module
sbt	Any recent sbt launcher; each module pins its own version (1.9.8 or 1.11.7) and sbt fetches it	Building each module
Apache Spark	3.5.9	Running a real job via spark-submit
Node.js	20	The results web UI (web/)
Python	3.9 or later, with apache-beam	Only for Beam Python pipelines

You don’t need to install these by hand if you use Codespaces — its dev container provisions all of them automatically.

See Engine Version Support for the full range of Spark 3.5.x patches and Beam releases actually verified, not just these defaults. If you only need Beam, you can skip Spark: see Use Invaract with Apache Beam or Beam Python pipelines.

Set up

Local machine

1. Install JDK 21, then verify:

   java -version

2. Install sbt via Coursier:

   curl -fsSL https://github.com/coursier/coursier/releases/download/v2.1.9/cs-x86_64-pc-linux.gz \
     | gzip -d > cs
   chmod +x cs
   ./cs install sbt

3. Install Apache Spark 3.5.9 and put spark-submit on your PATH:

   SPARK_VERSION="3.5.9"
   HADOOP_VERSION="3"
   wget "https://archive.apache.org/dist/spark/spark-${SPARK_VERSION}/spark-${SPARK_VERSION}-bin-hadoop${HADOOP_VERSION}.tgz"
   tar -xzf "spark-${SPARK_VERSION}-bin-hadoop${HADOOP_VERSION}.tgz" -C /opt
   mv "/opt/spark-${SPARK_VERSION}-bin-hadoop${HADOOP_VERSION}" /opt/spark
   export SPARK_HOME="/opt/spark"
   export PATH="$SPARK_HOME/bin:$PATH"

4. Install Node.js 20 (only needed for the results web UI).

5. Clone the repository:

   git clone https://github.com/mlltx/Invaract.git
   cd Invaract

JDK 17+ note

Spark reflectively accesses JDK-internal classes that JDK 17+‘s module system closes by default. Running via spark-submit (the primary path used throughout these docs) already sets the required --add-opens flags for you. If you invoke a module’s tests directly with sbt test and hit an InaccessibleObjectException, see Troubleshooting.

GitHub Codespaces

1. Clone the repository and open it in GitHub Codespaces:

   git clone https://github.com/mlltx/Invaract.git

   Open the repository on GitHub and choose Code → Codespaces → Create codespace.

2. Wait for the dev container to provision. .devcontainer/post-create.sh installs sbt, Apache Spark 3.5.9, and sets SPARK_HOME — about 5 minutes on first boot, and nothing further to configure.

3. Confirm the toolchain is ready:

   java -version
   sbt --version
   spark-submit --version

Next step

Continue to Quick Start to build every module and run a real, contract-verified Spark job.