conda-gdal-r-python

A minimal, conda-forge-based Docker image for cross-language R + Python geospatial work. Companion to gdal-r-ci — a different audience, the same underlying goal.

ghcr.io/hypertidy/conda-gdal-r-python:latest

What's in the image

Layer	Packages
Foundation	gdal · geos · proj · python ≥ 3.12 · r-base ≥ 4.5
Python geo	rasterio · pyogrio · fiona
R geo	r-terra · r-sf · r-gdalraster · r-vapour
Bridge	r-reticulate

All packages come from conda-forge. No apt-installed GDAL, no system R, no venv on top — one source of truth for every shared library.

RETICULATE_PYTHON is pinned to /opt/conda/bin/python so library(reticulate) finds the conda Python without extra configuration.

Quick start

docker pull ghcr.io/hypertidy/conda-gdal-r-python:latest
docker run --rm -it ghcr.io/hypertidy/conda-gdal-r-python

Run the sanity tests inside a container:

docker run --rm \
  -v "$PWD/local-test.sh:/local-test.sh" \
  ghcr.io/hypertidy/conda-gdal-r-python \
  bash /local-test.sh

Rebuild schedule

The image is rebuilt weekly (Mondays) to pick up whatever conda-forge ships. It is also rebuilt on any push that touches Dockerfile or environment.yml.

What this is not

• Not a replacement for Pangeo — Pangeo already serves the broader Python ecosystem. This image is specifically about R–Python interop on conda-forge.
• Not a kitchen-sink image. If you need xarray, zarr, dask, arrow, targets, etc., install them into a layer on top.
• Not gdal-r-ci. That image builds GDAL from source against a pinned master branch and targets a different audience (source-build, bleeding-edge). This image uses whatever conda-forge ships today.

Relationship to gdal-r-ci

	gdal-r-ci	conda-gdal-r-python
GDAL source	built from source (master / release)	conda-forge package
Audience	R+GDAL source-build users	conda-forge users
Versioning	:latest (release) + :dev (master)	:latest only
Purpose	bleeding-edge canary	conda-forge integration canary

License

MIT