4 releases 4 tags

RSS feed

• v1.1.2 70f81caf47

  Compare

  v1.1.2 Stable

  rbrooks released this 2026-09-26 10:41:59 +00:00 | 35 commits to main since this release

  Fixed

  • The published image could not run a scheduler job or push to the lights
    (#173). The Dockerfile installed from backend/requirements.txt, a
    hand-kept copy of pyproject.toml's dependencies that stopped being kept:
    prometheus-client (#15), defusedxml (#18) and zeroconf (#50) were
    added to pyproject.toml and never to it. Every scheduler job records
    itself through log_run, and every push through _try_push, and both
    import the metrics module — so in the 1.1.0 and 1.1.1 images the on, dim and
    off jobs all failed with ModuleNotFoundError: prometheus_client before
    doing anything, while the container started and passed its health check.
    Off-site S3 backup and controller discovery were broken the same way, and
    pytest and respx shipped in the runtime image instead.

    The image now installs from uv.lock: the build exports the locked runtime
    set with hashes and installs it in hash-checking mode, so what ships is
    exactly what the tests ran against, and there is no second list to drift —
    requirements.txt is gone. The weekly dependency audit, which had been
    auditing that same stale file, now audits the same export. And CI imports
    every app.* module inside the built image, which fails on the 1.1.1 image
    and passes on this one: nothing had ever checked that the image could import
    its own code, because the tests run from a full uv sync and the E2E never
    runs a job.

    Upgrading: nothing to do beyond pulling 1.1.2. Do not deploy 1.1.0 or
    1.1.1.

  Downloads

  • Source code (ZIP)
    0 downloads
  • Source code (TAR.GZ)
    0 downloads