geocompx
Julia-based images
- [ ] Base image with core spatial deps installed
- [x] Addition to other images for x-language work
Looks like there are still a few problems if you manage the Julia version via pixi: https://discourse.julialang.org/t/managing-julia-versions-using-pixi/116165/16
Interesting, seems there are a good few Julia projects here: https://pixi.sh/v0.16.1/Community/ some of which are by @evetion none of them use Docker or devcontainers as far as I can see though.
Pure Julia version doesn't need to use Pixi. rpyjl version: probably worthwhile but open to ideas!
Hmm...a pure Julia image would be pretty easy then! Is it easy to automatically update?
Yes, if you supply a Dockerfile we can handle the auto-deployment, just a julia.yml file in the .github/workflows folder should do it.
Interesting, seems there are a good few Julia projects here: https://pixi.sh/v0.16.1/Community/ some of which are by @evetion none of them use Docker or devcontainers as far as I can see though.
Correct, pixi is easier to install than docker and it does everything for us, using the tasks feature. The notable exception is testing our QGIS plugin, which is done in Docker.
I wonder to what extent pixi could replace Docker, with possible resource savings.
This is basically done in https://github.com/geocompx/docker/tree/master/rocker-rpyjl although in terms of image size... I think we can do better. See https://github.com/geocompx/docker/issues/110 for an approach starting with Python's official bookworm-slim image. Thoughts v. welcome @evetion, @asinghvi17 and others, is this a solved problem or is it still worth developing cross-language images here? My hypothesis: not solved and yes worth developing this!