versions.env

# Which cinepi-raw goes with which cinemate.
#
# These two programs talk to each other over a redis key contract, and nothing
# anywhere has ever recorded which revisions were tested together. The
# installer pins the sensor drivers (cinemate-modes) and libcamera (the cinemate
# branch), but left the two repositories that move most tracking whatever their
# default branch happened to hold that day -- so an install performed on two
# different days is not necessarily the same system.
#
# For scale: cinepi-raw's main and dev currently differ by 45 files and +7164
# lines, including four CONTROL_KEY_ macros of that shared contract.
#
# ---------------------------------------------------------------------------
# HOW TO USE THIS
#
# Both values are EMPTY by default, which reproduces exactly today's
# behaviour: clone the default branch of each. Nothing changes until you fill
# them in.
#
# Fill them in once you have an install you trust:
#
#   1. Do a clean install and verify the camera reaches ready and records.
#   2. On that machine:
#        git -C ~/cinemate    rev-parse HEAD
#        git -C ~/cinepi-raw  rev-parse HEAD
#   3. Put those two values below, with the date and what you verified.
#   4. Commit it. That is now a pairing someone else can reproduce.
#
# Update it whenever a change to one repo needs a matching change in the other
# -- a new CONTROL_KEY_, a changed command-line flag, a new sensor mode.
#
# Any value here can still be overridden from the environment, as before:
#   CINEPI_RAW_REPO_REF=some-branch ./cinemate-install.sh
# ---------------------------------------------------------------------------

# Last verified pairing: (none recorded yet)
# Verified on:           --
# Verified by:           --

CINEMATE_REPO_REF="${CINEMATE_REPO_REF:-}"
CINEPI_RAW_REPO_REF="${CINEPI_RAW_REPO_REF:-}"

# For reference, the revisions this remediation work was written against:
#   cinemate    dev  02b5a395922f8ed63d35498562142fe1ac08abb3
#   cinepi-raw  dev  ea96f2d3da98511df3a2ba001fb54433445017d4
# Neither has been validated on hardware, so they are recorded here as a
# starting point rather than set above as a known-good pairing.