arch(light,cover): default to native light transitions and extract bus-truthful cover kinematics strategies #730

Architectural Assessment & Severity

• Critique Point: Sending multiple commands will fail #1 — Software-stepped fades and virtual cover positioning
• Severity: P0 (Critical — Rank Sending multiple commands will fail #1)
• Related Issues: Covers: end a timed run at the stop status, not the stop write (#466) #705, Covers: per-cover motor start-up floor instead of one 0.4 s constant (#466) #704, Send queue: let a cover's stop frame overtake queued frames #645, feat(cover): support venetian blind slat / louvre tilt positioning (WHO 2 Dimension 11) #492, Community Call for Real-World Gateway Traces: Help Us Eliminate Hardware Blind Spots! 📡 #466, feat(cover,lock): impulse garage/gate control, multi-factor safety approval, and intercom events #667, RFC: timed covers, what I measured on a MyHOMEServer1 (latencies, frame order, non-linear position) #302

Problem Statement

1. Software Fade Engine (custom_components/myhome/light_fade.py, light.py, const.py)

• DEFAULT_TRANSITION_MODE is currently set to TRANSITION_MODE_SOFTWARE ("software_stepped") at custom_components/myhome/const.py:175 (and TRANSITION_MODE_AUTO also resolves to TRANSITION_MODE_SOFTWARE at custom_components/myhome/light.py:603-604).
• Consequently, every light.turn_on(..., transition=T) call by default spawns a software loop sending up to 25 instant-brightness frames (*#1*WHERE*#1*level*0##) at 0.3s intervals (SOFTWARE_TRANSITION_STEP_INTERVAL = 0.3, light_fade.py:159-188) at Tier-0 command priority.
• On a shared, low-bandwidth SCS bus, this floods the gateway during multi-light scenes or adaptive_lighting updates, fights actuators that natively support hardware transition speeds (*#1*WHERE*#1*level*speed##), can reorder steps when command_worker_count > 1 (light_fade.py:132-141), and requires an optimistic echo-suppression rule (abs(reported - current_opt) >= 10 at light.py:1100-1108) that ignores real bus state during active fades.

2. Virtual Travel-Time Covers (custom_components/myhome/cover.py, cover_motion.py, cover_calibration.py)

• Standard WHO=2 actuators (F411U2, F411/4, LN4660M2) do not report percentage position on the bus (only F430 / Dimension 10 *#2*WHERE*10## actuators do).
• However, MyHOMECover (cover.py:195-1353) forces virtual travel-time estimation (DEFAULT_TRAVEL_TIME = 25 at const.py:105) and a fabricated startup position of _attr_current_cover_position = 50 (cover.py:280) onto every uncalibrated standard cover by default.
• All cover behaviors—native hardware position (advanced=True), virtual travel-time dead-reckoning, interactive bus calibration (cover_calibration.py), echo-window tracking (ECHO_WINDOW = 2.0), delivery-confirmation futures, motor start-up lag (MOTOR_START_DELAY), stop-status re-anchoring (Covers: end a timed run at the stop status, not the stop write (#466) #705), per-cover motor pulse floor (Covers: per-cover motor start-up floor instead of one 0.4 s constant (#466) #704), deadband protection (Community Call for Real-World Gateway Traces: Help Us Eliminate Hardware Blind Spots! 📡 #466), Venetian slat tilt (feat(cover): support venetian blind slat / louvre tilt positioning (WHO 2 Dimension 11) #492), and impulse gate/garage timers (feat(cover,lock): impulse garage/gate control, multi-factor safety approval, and intercom events #667)—are entangled in MyHOMECover (1,353 lines on base, growing past 2,200+ lines across 5 concurrent cover branches).

Proposed Solution (PR Scope)

Part A — Lights (const.py, light.py, light_fade.py)

1. Default to Native Hardware Transitions:
   • Change DEFAULT_TRANSITION_MODE in custom_components/myhome/const.py:175 from TRANSITION_MODE_SOFTWARE to TRANSITION_MODE_NATIVE ("native").
   • Map TRANSITION_MODE_AUTO in custom_components/myhome/light.py:603-604 to TRANSITION_MODE_NATIVE.
2. Demote & Isolate Opt-In Software Fades:
   • Keep TRANSITION_MODE_SOFTWARE strictly opt-in for legacy dimmers that ignore hardware speed parameters.
   • Enqueue intermediate software fade steps at a lower priority tier (_TIER_FADE_STEPS = 1, below interactive Tier-0 commands) so fades never delay wall-switch or cover STOP commands.
   • Remove the <10% optimistic echo suppression during native transitions, and immediately cancel an opt-in software fade if an external physical switch event arrives.

Part B — Covers (cover.py, cover_strategies/)

1. Extract Pluggable CoverKinematicsStrategy Classes (custom_components/myhome/cover_strategies/):
   • SimpleUpDownStopStrategy (new default for uncalibrated/unconfigured standard covers): Exposes OPEN | CLOSE | STOP only, reports is_opening / is_closing / is_closed truthfully from bus frames (*2*1/2/0*WHERE##), sets current_cover_position = None (no fabricated 50% startup state), and runs zero background stopwatch timers.
   • NativePositionStrategy: For advanced=True actuators or any cover auto-promoted on receiving a Dimension 10 (*#2*WHERE*10##) frame.
   • TimedCurtainStrategy: Enabled only when travel_time is explicitly configured or calibration data exists in CONF_COVER_TRAVEL_TIMES. Encapsulates dead-reckoning, min_motor_pulse (Covers: per-cover motor start-up floor instead of one 0.4 s constant (#466) #704), cover_deadband_pct (Community Call for Real-World Gateway Traces: Help Us Eliminate Hardware Blind Spots! 📡 #466), and stop-status re-anchoring (Covers: end a timed run at the stop status, not the stop write (#466) #705).
   • VenetianTiltStrategy: Encapsulates slat tilt pulse math (feat(cover): support venetian blind slat / louvre tilt positioning (WHO 2 Dimension 11) #492).
   • ImpulseDoorStrategy: Encapsulates gate/garage impulse timers (feat(cover,lock): impulse garage/gate control, multi-factor safety approval, and intercom events #667).
2. Slim MyHOMECover (cover.py):
   • Reduce MyHOMECover to a thin Home Assistant CoverEntity adapter (~350 lines) that delegates kinematics to self._strategy.

Verification Plan

• Unit tests in tests/test_cover_strategies.py covering SimpleUpDownStopStrategy (uncalibrated default, no fake 50%), NativePositionStrategy, TimedCurtainStrategy, VenetianTiltStrategy, and ImpulseDoorStrategy.
• Unit tests in tests/test_light.py verifying DEFAULT_TRANSITION_MODE == "native" sends a single native transition command instead of stepped frames.
• Full WSL test suite (pytest tests -q --cov=custom_components/myhome), ruff check, and scripts/verify_ha_standards.py.