[GreenGrassBlueOcean]

Follow-up to #739 addressing the two review items from @Interstellar0verdrive.

---

1. What is the problem?

1. A keypad or bus run that hits the physical limit switch reported "stop" instead of "end_stop" (custom_components/myhome/cover.py):
   • On the OpenWebNet bus, when a standard timed shutter actuator (BT-F411/2, F411/4, LN4672M2) reaches its limit switch / hardware relay cutoff at the end of a keypad run (*2*1*WHERE## or *2*2*WHERE##), the actuator emits *2*0*WHERE## (WHAT=0, identical on the wire to a manual keypad stop).
   • Because MyHOMECover._freeze_position(at) unconditionally passed "stop" to self._strategy.on_run_ended(self, "stop", frozen, at) without checking whether the run had lasted at least the full directional travel time (at - run_started_at >= _travel_for(is_opening)), a full-travel keypad run reported "stop" instead of "end_stop" and never called self._strategy.on_position_confirmed(self, frozen, "end_stop").
   • For stateful strategies that leave position None until an endpoint is confirmed (such as PiecewiseGeometryStrategy(known=False) in tests/test_cover_strategy_contract.py and the calibrated geometry tracker in feat: add dedicated MyHOME administration panel #374), a full-travel keypad run left the position None (unknown) instead of confirming 100 or 0.
2. Memberless scope cover run timeout set 100/0 without confirming the position (custom_components/myhome/cover_scope.py):
   • Under WHO_2.pdf §3.0.1, general (WHERE="0"), area (WHERE="1".."9", "00", "100"), and group (WHERE="#1"..) commands never receive a stop frame for the scope WHERE itself. When a MyHOMEScopeCover has no configured member entities, _watch_run() schedules _end_run after self._travel_for(moving == "open").
   • _end_run called self._freeze_position(time.monotonic()) and then overwrote self._attr_current_cover_position = self._start_position = 100 if opening else 0 without advancing the freeze timestamp to _run_started_at + travel or calling self._strategy.on_position_confirmed(self, end, "end_stop").

---

2. How is it solved?

1. Detect full-travel runs in MyHOMECover._freeze_position(at) (custom_components/myhome/cover.py):
   • Computes end_stop when not self._strategy.assumed_state, run_started_at is not None, the cover is opening or closing, (travel := self._travel_for(is_opening)) > 0, and at - run_started_at >= travel.
   • When end_stop is True:
     • Sets frozen = 100 if is_opening else 0 (even when position_after would otherwise return None because the strategy had no prior known position).
     • Calls self._strategy.on_run_ended(self, "end_stop", frozen, at) followed by self._strategy.on_position_confirmed(self, frozen, "end_stop").
2. Confirm the endpoint in MyHOMEScopeCover._end_run() (custom_components/myhome/cover_scope.py):
   • Advances the freeze timestamp at to max(time.monotonic(), self._run_started_at + travel) when _run_started_at is set (accounting for any queue delay between scheduling _watch_run and anchoring the motor write), so _freeze_position(at) reports ("end_stop", end) and confirms end (100 or 0).
   • If _end_run fires when no run anchor was set (not confirmed), explicitly invokes self._strategy.on_position_confirmed(self, end, "end_stop").
3. Expand contract test coverage (tests/test_cover_strategy_contract.py):
   • Verifies that the 13 s closing keypad run in test_piecewise_strategy_drives_cover_without_cover_py_changes records ("end_stop", 0) and ("confirmed", 0, "end_stop").
   • Verifies in test_position_unknown_until_a_full_run that a full-travel keypad run (*2*2*21## $\rightarrow$ 13.0 s $\rightarrow$ *2*0*21##) and a memberless MyHOMEScopeCover run timeout (*2*1*#7## $\rightarrow$ _end_run) both transition PiecewiseGeometryStrategy(known=False) to known = True and confirm the endpoint.

---

3. State Diagram

stateDiagram-v2
    direction TB
    [*] --> UnknownPosition: Startup / Restart\n(strategy.known == False, position == None)

    state "Position Unknown (None)" as UnknownPosition
    state "Motor Running (Anchored at run_started_at)" as Running
    state "Partial Stop (elapsed < travel)" as PartialStop
    state "Confirmed End Stop (0% or 100%)" as ConfirmedEndStop
    state "Known Intermediate Position (0..100%)" as KnownPosition

    UnknownPosition --> Running: Keypad / Bus Move (*2*1*W## or *2*2*W##)\non_run_started(direction, start_position, at)
    KnownPosition --> Running: Move or Set Position\non_run_started(direction, start_position, at)

    Running --> PartialStop: *2*0*W## or async_stop_cover()\nwhere at - run_started_at < travel_for(opening)
    PartialStop --> UnknownPosition: strategy.known == False\non_run_ended("stop", None, at)
    PartialStop --> KnownPosition: strategy.known == True\non_run_ended("stop", frozen, at)

    Running --> ConfirmedEndStop: 1) *2*0*W## after at - run_started_at >= travel_for(opening)\n2) _async_run_to_end() settle timer (0% / 100%)\n3) MyHOMEScopeCover._end_run() timeout
    ConfirmedEndStop --> KnownPosition: on_run_ended("end_stop", 100|0, at)\non_position_confirmed(100|0, "end_stop") -> strategy.known = True

Loading

---

4. Where are the traces (real ones)?

Hardware Gateway	Firmware / Profile	Real Trace / Log File	Observed Shutter End-of-Run Behavior
BTicino F461	Firmware 2.0.11 + BT-F411/2 (WHERE="05", 5 s hardware cutoff)	tests/fixtures/traces/f461_live_2026-10-18/myhome_trace_F461_live_2026-10-18.json	In run2_shutter_05_travel_and_reversal, after *2*2*05## starts a full closing run, the physical BT-F411/2 actuator emits a standard *2*0*05## stop frame on evt (dt = 5.0 s after motor start) when its hardware travel timer expires at the limit switch.
BTicino MyHOMEServer1	Firmware 2.38.9 (028206) + Timed Shutters (WHERE="21", "31")	tests/fixtures/traces/myhomeserver1_timed_cover_echo_window.json & tests/test_issue_445_myhomeserver1_covers.py	Relays *2*1000#0*W##, *2*0*W##, *2*1000#1*W##, *2*1*W## on motor start (~0.55 s) and emits *2*0*W## when the actuator's relay drops out at the end of the run (~60 s uncalibrated cutoff or calibrated travel time).
BTicino MH201	Firmware 1.0.x + Group/Area Scope Shutters (#433)	tests/test_issue_433_cover_scope.py	Scope commands (*2*1*0##, *2*1*1##, *2*1*#7##) never emit a stop frame for the scope WHERE (WHO_2.pdf §3.0.1); only individual actuators emit *2*0*<point>## at the end of travel, so a memberless scope cover settles via _end_run after _travel_for(opening).

---

5. What does the Firmware Oracle / Encyclopedia say?

1. OpenWebNet Encyclopedia (functional/who-2-automation/what.md & WHO_2.pdf §3.0.1):
   • Standard WHO=2 basic movement defines only WHAT=0 (Stop), WHAT=1 (Up), and WHAT=2 (Down). There is no distinct WHAT code for a hardware limit-switch stop on non-advanced shutters: when a point-to-point actuator reaches the end of its run, it transmits *2*0*WHERE## (WHAT=0).
   • Under WHO_2.pdf §3.0.1, a broadcast Up or Down to GEN (0), A (area), or GR (#G) terminates with individual *2*0*<point>## frames from each physical automation object moved ("as many frames as automation objects"), and never with *2*0*0## / *2*0*A## / *2*0*#G## for the scope address itself.
2. Empirical Firmware Oracle (openwebnet-mcp — lookup_firmware_verdict):
   • Across the emulated gateway fleet (F453, F453AV, F454, F459, F460, F461, H4684, MH200, MH200N, MH202, MyHomeServer1), *2*1*31## / *2*2*31## and *2*0*31## are the sole event-stream state transitions for standard WHO=2 shutters (F460, F461, and MyHomeServer1 additionally emit *2*1000#1*31## / *2*1000#0*31## translation frames before the actuator status frame). Distinguishing a manual mid-travel stop from a limit-switch end stop on a standard shutter is therefore a function of whether the motor ran for at least travel_for_direction(opening).

---

Verification

• pytest tests/test_cover_strategy_contract.py tests/test_issue_433_cover_scope.py tests/test_component_cover.py tests/test_cover_strategies.py

• mypy custom_components/myhome/cover.py custom_components/myhome/cover_scope.py custom_components/myhome/cover_strategies (Success: no issues found in 8 source files)

• ruff check custom_components tests scripts

• scripts/verify_ha_standards.py & scripts/sync_documentation.py --check

• V2 Architecture: New entities implement handle_event() and do not manually subscribe via async_dispatcher_connect.

[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!