Installation

If you need help, reach out on GitHub.

Atoti can be installed either as a Python package or as a Conda package.

Python package (recommended)

Atoti and its optional dependencies (such as the JupyterLab extension) are available on PyPI and can thus be installed with package managers such as uv:

uv add "atoti[jupyterlab]"

or pip:

pip install "atoti[jupyterlab]"

Advanced

ActiveViam repository

Installing Atoti Python SDK packages with PyPI is the easiest way but they can also be downloaded from ActiveViam’s repository by configuring https://activeviam.jfrog.io/artifactory/api/pypi/pypi/simple/ as an extra-index-url.

Client only

Tip

Projects using Atoti Python SDK exclusively to connect to already running Atoti sessions (i.e. never starting new ones) do not need to install the atoti package and can instead install atoti-client.

atoti-client and its extras do not require a Java runtime and do not embed any JARs so they take up significantly less disk space.

uv add atoti-client

Or, with the JupyterLab extension extra:

uv add "atoti-client[jupyterlab]"

Custom JDK

Installing atoti will install jdk4py: a Python package embedding a JDK including all the Java modules required to run Atoti.

Projects willing to use their own JDK can skip the installation of jdk4py:

uv add atoti-client atoti-server

This requires the JAVA_HOME environment variable to point to a JDK >=21.

Conda package (discouraged)

Install conda with an installer such as Miniforge .

Add the Atoti channel:

conda config --add channels https://activeviam.jfrog.io/artifactory/api/conda/conda

Create a new Conda environment:

conda create --name atoti

Activate it:

conda activate atoti

Install Atoti and its JupyterLab extension:

conda install atoti atoti-jupyterlab

Offline

Tools such as pip download or conda pack can be used to facilitate offline installation.