Repository navigation
Conversation
|
@tbelz thanks for the MR. One comment: The Dockerfile rebuilds all dependencies every time there is a change the code. Docker builds an image step by step, and it caches each step. If nothing a step depends on has changed, Docker skips it and reuses the cached result. But the moment one step's inputs change, that step and every step after it must re-run. What the PR does: In the builder stage, the steps are in this order:
The problem: Because the source code is copied before the install step, editing even one line of jmcp.py invalidates step 2, which forces step 3 to re-run — meaning a full re-download and re-compile of every dependency, even though the dependencies didn't change at all. The fix: The install step only needs pyproject.toml and uv.lock — it never reads the source code (that's what the --no-install-project flag means). So copy only those two files into the builder stage, run the install, and copy the source code into the final image directly from your project folder. Then a code change only re-runs the cheap "copy files" step, and the expensive dependency install stays cached. Can you take care of above ? |
Summary
Cross-repository design authority:
https://github.com/tbelz/nita/tree/codex/container-security-baseline/openspec/changes/archive/2026-08-05-enforce-container-security-baseline
Tracks Juniper/nita#75. The supported boundary remains a locally operated trusted lab, not an internet-facing or multi-tenant deployment.
Validation