Publish a Python Wheel to GCP Artifact Registry with Poetry

a year ago

The author needed a runnable Python file instead of a Docker image to interact directly with the machine and GPU drivers.
Poetry was used to build a wheel file for the Python project, with configurations in pyproject.toml to include the main directory.
A Python registry was created in GCP's Artifact Registry to store the wheel file, requiring setup for both pushing and pulling the file.
Dynamic versioning was implemented in pyproject.toml to avoid overriding files in the registry, using a timestamp-based version number.
The VM in GCP was configured with the necessary permissions and settings to pull the wheel file from the Artifact Registry.
The project was installed using pip, which checks both the public registry and the custom GCP registry for dependencies.
The application was run directly using Python, bypassing the need for Docker, though manual dependency management in the VM was noted as a drawback.

Hasty Briefsbeta