🌍 Here Be Dragons 🐲

A helpful Docker image for bleeding-edge spatial data analysis in Python.

Tricky dependencies are built from source so you can run the latest Python geoprocessing libraries that depend on Arrow, GDAL, GEOS, and PROJ.

The images are built on top of the most recent Ubuntu release (ubuntu:rolling) with optimized Python builds. The Python package managers uv and Pixi come pre-installed.

By default, running the Docker images launches a marimo notebook.

Tip

You will need to install Docker. On Windows, running Docker Desktop with Windows Subsystem for Linux (WSL) is recommended.

🐍 Latest

This Docker image is the stable version intended for production work with common geospatial libraries in the latest version of Python.

To pull and start the container in your terminal:

docker run --name here-be-dragons-latest -p 8080:8080 ghcr.io/corbel-spatial/here-be-dragons:latest

To update and restart the stopped container:

docker pull ghcr.io/corbel-spatial/here-be-dragons:latest
docker start here-be-dragons-latest
docker attach here-be-dragons-latest

🔥 Next

This Docker image is the experimental version intended for testing the latest pre-release version of Python against Arrow, GDAL, GEOS, and PROJ. Failing builds might indicate that issues are on the horizon. Some extra packages (mapclassify, etc.) may not be included in this image.

docker run --name here-be-dragons-next -p 8080:8080 ghcr.io/corbel-spatial/here-be-dragons:next

⚠️ TODO

• Scheduled automatic update of versions.env to find new releases/tags
• Add latest-dev and next-dev variants that keep all the build dependencies
• Add variants built with freethreaded Python (--disable-gil)
• Create a benchmark suite and check new releases for performance regressions
• Add more packages upon request - please open an issue!

📚 References

Project	Source	Downstream Packages
Arrow	https://github.com/apache/arrow	geoarrow-pyarrow
GDAL	https://github.com/OSGeo/gdal	Python bindings
GEOS	https://github.com/libgeos/geos	Shapely
PROJ	https://github.com/OSGeo/PROJ	Pyproj
Python (release schedule)	https://github.com/python/cpython	uv, Pixi
Ubuntu	https://hub.docker.com/_/ubuntu	n/a