Standalone mode ocp-cad-viewer with Docker

This is Docker packaging for https://github.com/bernhard-42/vscode-ocp-cad-viewer .

The advantage of Docker is that it keeps the Flask based standalone web viewer in separated environment.

You will still have to of course pip install ocp_vscode and build123d into the venvs where you work, but those packages are pure Python, without dependencies like xrender or libgl.

Starting the docker container and opening the viewer

First of all, you have to run the container that provides the web server:

docker run -it --rm -p 3939:3939 ghcr.io/nilcons/cad-viewer

Once, this is running, you can visit localhost:3939.

Alternatively you can make a full window app out of it like this:

#!/bin/bash

set -euo pipefail

docker run --name cadview --detach -p 3939:3939 --rm ghcr.io/nilcons/cad-viewer --reset_camera keep || true
google-chrome --user-data-dir=$HOME/ocp-dedicated-chrome-data-dir --app=http://localhost:3939/

The docker run part's errors are ignored, so if the container is already running, the chrome window still opens.

This example also shows that you can pass in the usual vscode-ocp-cad-viewer command line flags at the end of the docker run command. To get a list of supported flags you can run docker run --rm ghcr.io/nilcons/cad-viewer --help.

Sending data to the viewer for display

Once your browser is connected, you can send data to it as normal with ocp_vscode's show functions:

from build123d import *
from ocp_vscode import *

b = Box(1,2,3)
show(b)

Of course, in the venv where you are running this, build123d and ocp_vscode has to be pip installed.