[GreenGrassBlueOcean]

What is the problem

In Issue #746, @wave68runner reported that switch.luce_specchio_bagno (WHERE=04, Area "00", PL=04, configured on a BTicino MH200N gateway as a WHO=1 switch) updates normally when toggled directly from the wall button (*1*1*04## / *1*0*04##) or from Home Assistant, but remains stuck ON in Home Assistant when turned off via the physical General OFF wall button.

Analyzing the 117-frame OpenWebNet trace attached to #746 (myhome_trace_MH200N_all_2026-10-17T14-28-20.json) reveals the exact root cause:

1. When the physical General OFF button is pressed (*15*01*81##), MH200N executes Scenario 6 (*17*1*6## -> *17*2*6##).
2. Scenario 6 emits both the broadcast General OFF frame (*1*0*0##) and individual point-to-point commands for a subset of actuators in Area "00" (01, 03, 09, 0011), without emitting a dedicated point-to-point frame for 04 (because 04 is already turned off by the *1*0*0## broadcast).
3. As of #732 (bus-truthful state), MyHOMELight and MyHOMESwitch entities intentionally ignore broadcast frames (WHERE="0" and area WHERE="00", "1".."9", "100") and rely on LightingResyncManager (custom_components/myhome/gateway_resync.py) to poll *#1*<area>## after the 0.5 s debounce window.
4. However, LightingResyncManager treated any single point-to-point echo in an area (any(a == raw_where for _, _, a in self._recent_ptp) for leading echoes, or the first trailing echo in handle_ptp_echo) as proof that the entire area had already reported its state:
   • Because 01, 03, 09, and 0011 echoed in Area "00" during Scenario 6, LightingResyncManager skipped/cancelled *#1*00## while still polling *#1*1## through *#1*5##.
   • Consequently, MH200N was never asked *#1*00##, never emitted *1*0*04##, and switch 04 remained stuck ON in Home Assistant.

(Note: Open PR #740 partially touched _area_fully_echoed, but gated the subset check on if len(known) > 1: — which still fails when only 1 entity in Area "00" is registered in Home Assistant while unregistered siblings echo on the bus — and did not filter _recent_ptp by latest per-entity polarity, allowing a prior opposite-polarity transition such as turning 04 ON < 1.5 s before *1*0*0## to falsely satisfy _area_fully_echoed.)

---

How is it solved

1. Per-entity area registry map (_known_area_map & known_area_entities):
   • LightingResyncManager now maps each area code ("00", "1".."9", "100") to the normalized set of WHERE addresses (normalize_where(address.clean_where)) registered under light or switch for the gateway's config_entry.
2. Completeness check across all registered area entities (_area_fully_echoed):
   • Whenever known_area_entities(area) is non-empty (if known: return known.issubset(echoed_wheres)), an area resync sweep (*#1*<area>##) is only skipped (leading window) or cancelled (trailing debounce window) once every registered light/switch entity in that area has emitted a point-to-point echo.
   • Even if only a single entity (04) is registered in Area "00", echoes from unregistered actuators (01, 03, 09, 0011) cannot cancel *#1*00## until 04 itself echoes.
3. Latest per-entity polarity matching (_ptp_polarities & _resync_area_echoes):
   • _leading_echoes_for_area() evaluates the latest point-to-point polarity recorded for each entity in _recent_ptp within RESYNC_LEADING_WINDOW_S (1.5 s) and only counts echoes whose polarity matches the broadcast (msg_is_on), preventing an ON transition (*1*1*04##) right before General OFF (*1*0*0##) from masquerading as a leading OFF echo.
   • Matching leading echoes seed self._resync_area_echoes[area], which then accumulates trailing echoes during the 0.5 s debounce window and cancels the timer only if _area_fully_echoed(area, echoed) becomes True.

---

How is it tested

• Authentic MH200N trace replay (tests/test_issue_746_mh200n_gen_off_switch_trace.py):
  • Ingests @wave68runner's 117-frame MH200N capture (tests/fixtures/traces/issue_746/myhome_trace_MH200N_all_2026-10-17T14-28-20.json).
  • Replays the wall switch turn-on (*1*1000#1*04## -> *1*1*04##) followed by the CEN pushbutton (*15*01*81##), MH200N Scenario 6 (*17*1*6## -> *17*2*6##), partial Area "00" echoes (01, 03, 09, 0011), and General OFF (*1*0*0##).
  • Verifies that Area "00" remains armed in _resync_timers, dispatches *#1*00## alongside *#1*1##..*#1*5##, and transitions switch_04 to OFF upon receiving *1*0*04##.
• Unit & edge-case tests (tests/test_issue_368_resync.py):
  • test_partial_area_echoes_do_not_skip_or_cancel_area_sweep: verifies partial leading + trailing echoes in Area "00" (01, 03, 09, 0011 without 04) still emit *#1*00##, whereas full echoes (01, 03, 04, 09, 0011) cancel the timer without polling.
  • test_single_registered_entity_in_area_not_cancelled_by_unregistered_sibling_echo: verifies the len(known) == 1 edge case where only 04 is registered in Area "00" and unregistered 01 echoes.
  • test_opposite_polarity_leading_echo_does_not_skip_area_sweep: verifies turning 04 ON (*1*1*04##) within 1.5 s before General OFF (*1*0*0##) does not suppress *#1*00##.
• Coverage & Standards:
  • 100.0% line coverage on custom_components/myhome/gateway_resync.py (162 Stmts, 0 Miss).
  • ruff check, scripts/verify_ha_standards.py, and scripts/sync_documentation.py --check all pass with zero warnings.

---

State Diagram

stateDiagram-v2
    [*] --> Idle

    Idle --> LeadingWindowCheck : RX Broadcast (*1*0*0## or *1*0*00##)
    state LeadingWindowCheck {
        [*] --> CollectLeadingEchoes
        CollectLeadingEchoes : Filter _recent_ptp (last 1.5s) for Area "00"\nusing latest polarity per entity == msg.is_on
        CollectLeadingEchoes --> CheckLeadingComplete : known_area_entities("00") = {"01","03","04","09","0011"}
    }

    CheckLeadingComplete --> Idle : known ⊆ leading_echoes\n(All entities already echoed -> skip sweep)
    CheckLeadingComplete --> DebounceArmed : known ⊈ leading_echoes\n(e.g. "04" missing -> seed _resync_area_echoes["00"])

    state DebounceArmed {
        [*] --> WaitingForTrailingEchoes : Arm 0.5s timer for Area "00"
        WaitingForTrailingEchoes --> AccumulateEcho : RX PTP in Area "00"\n(*1*0*01##, *1*0*03##, *1*0*09##, *1*0*0011##)
        AccumulateEcho --> WaitingForTrailingEchoes : echoed = {"01","03","09","0011"}\n("04" still missing -> keep timer armed)
    }

    DebounceArmed --> Idle : RX *1*0*04## before 0.5s\n(known ⊆ echoed -> cancel timer)
    DebounceArmed --> PollArea00 : 0.5s timer expires\n("04" never echoed)

    PollArea00 : TX *#1*00## (OWNLightingCommand.status("00"))
    PollArea00 --> BusStatusReply : MH200N queries SCS Area 00
    BusStatusReply : RX *1*0*04## from MH200N
    BusStatusReply --> SwitchUpdated : Router dispatches to MyHOMESwitch("04")\nswitch.luce_specchio_bagno -> OFF
    SwitchUpdated --> [*]

Loading

---

What does the Firmware Oracle / Encyclopedia say

1. Firmware Oracle (openwebnet-mcp — lookup_firmware_verdict & parse_and_validate_frame)

Querying the emulated firmware oracle for MH200N (010108, bt_fls) and comparing across the gateway matrix confirms:

Frame	Purpose	MH200N (010108) Verdict	SCS Bus Telegram Emitted (OUT (Bus))	Cross-Gateway Support
*#1*00##	Area 00 status request	ACK (*#*1##) -> OUT (Bus)	24 30 34 42 33 30 30 31 35 30 30 0d ($04B3001500\r)	ACK + Bus TX on MH200N, F453AV, F454, F459, F460, F461, H4684, MH202, MyHomeServer1 (NACKed only on legacy MH200 020102)
*#1*2##	Area 2 status request	ACK (*#*1##) -> OUT (Bus)	24 30 34 42 33 30 32 31 35 30 30 0d ($04B3021500\r)	Identical $04B3<AA>1500\r SCS area-query telegram structure to *#1*00##
*#1*02##	Point-to-point status request (A=00, PL=02)	ACK (*#*1##) -> OUT (Bus)	24 30 34 42 30 30 32 31 35 30 30 0d ($04B0021500\r)	ACK + Bus TX on all gateways (A=00, PL=01–09 uses 2-digit 01–09 addressing)
*1*1*02## / *1*0*02##	Point-to-point ON / OFF (A=00, PL=02)	ACK (*#*1##) -> OUT (Bus)	24 30 34 42 30 30 32 31 32 30 30 0d / ...31 0d	Confirms WHERE=04 (A=00, PL=04) is a standard 2-digit Area 00 point-to-point address

2. OpenWebNet Encyclopedia Machine KB (WHO=1 Lighting & WHO=17 MH200N Scenarios)

• Area 00 Addressing (WHO=1):
  • In WHO=1, Area 00 point-to-point addresses with PL 1..9 are encoded as two digits 01..09 (e.g. WHERE=04), and PL 10..15 are encoded as four digits 0010..0015 (e.g. WHERE=0011), while the Area 00 broadcast/status target is WHERE=00 (*#1*00##).
• Silent Relay Actuation on General / Area Broadcasts (*1*0*0## / *1*0*<A>##):
  • Physical SCS relays actuated by a General (WHERE=0) or Area (WHERE=00, 1..9, 100) broadcast switch their load simultaneously without transmitting individual point-to-point *1*0*<WHERE>## status frames to avoid colliding on the 9600-baud SCS bus.
  • When an MH200N scenario (WHO=17, *17*1*6##) runs a mixed sequence containing both *1*0*0## and explicit point-to-point commands for specific actuators (01, 03, 09, 0011), only the explicitly addressed actuators emit point-to-point echoes; the remaining actuators in Area "00" (such as WHERE=04) only report their new OFF state when polled via *#1*00##.

Closes #746
Refs #748

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!