Moving from Conda Propagation to the Python Environment Provider node
The Conda Environment Propagation node relies on Conda to manage and bundle environments. It remains fully supported for existing workflows and specific infrastructure needs.
However, the new Python Environment Provider node offers a more modern approach. It uses Pixi under the hood, a fast, cross-platform package manager that uses a lock-file system to guarantee reproducibility across Windows, macOS, and Linux, and fully supports execution via KNIME Hub and the modern web-based UI.
How to migrate your workflow:
- Swap the nodes: Replace the "Conda Environment Propagation" node with the new "Python Environment Provider" node in your workflow.
- Configure packages: Open the configuration dialogue of the Python Environment Provider node. Instead of selecting an existing local Conda environment, explicitly add the Python packages (e.g.,
pytorch,plotly) your script requires. - Connect the ports: Connect the output port of the Python Environment Provider node to the flow variable or environment input port of your Python Script or Python View nodes.
- Execute: When executed, the node will resolve the dependencies and quickly build a portable, isolated environment. By default, this environment is temporary and scoped to the node execution, making it perfectly safe for sharing on KNIME Hub.