OpenCADC Base Images

This repository contains the layered Dockerfile definitions for the OpenCADC/CANFAR container ecosystem.

Architecture

The images follow a strictly layered, multi-stage Docker model to ensure consistency, reproducibility, and auditability:

1. Runtime Layer (Stage 1: runtime):
   • OS: Ubuntu 24.04 LTS (Digest Pinned)
   • Runtime: Micromamba + Python 3.13
   • Tools: uv, pixi (Version Pinned)
   • Focus: OS foundation and language runtime stability.
2. Science Layer (Stage 2: science):
   • Inherits from Runtime
   • Science Stack: cadcdata, cadctap, cadcutils, canfar, vos
   • Focus: Domain-specific tools and library ecosystem.
3. Terminal Image:
   • Inherits from Science
   • Adds: ttyd, starship (Pinned), tmux, htop, rclone, jq
4. Terminal OC Image:
   • Adds: OpenCode AI integration

Build Instructions

Because these images inherit from each other, they must be built in order:

# 1. Build the Base
docker build -t opencadc-base:latest ./base

# 2. Build the Standard Terminal
docker build -t opencadc-terminal:latest ./terminal

# 3. Build the Terminal with OpenCode
docker build -t opencadc-terminal-oc:latest ./terminal-oc

Configuration

• Ports: All terminal images expose port 5000 for ttyd.
• Theme: Uses the Gruvbox palette via Starship.
• Aliases: py -> python3, oc -> opencode, and standard navigation shortcuts.
• Persistence: Optimized for CANFAR Skaha. Conda environments, Pip packages (--user), UV cache, and Pixi homes are redirected to the user's persistent ${HOME} directory.
• Environment: Modern managers like uv and pixi are pre-configured with shell completions.