[gdluck]

Requested by A

Before: the #136 CHANGELOG entry said send_to_gateway refuses WHO 4 dimension 22 "with the contract above", which includes "kept in last_error", although that method creates and discards its session, so nothing is observable there besides the ERROR log and the None (or the raise with raise_on_error=True).

After: the entry describes send_to_gateway's refusal as it is, and the subject list reads "OWNCommandSession.send (and so AsyncTcpTransport.send and keepalive), and AsyncSerialTransport.send refuse it".

Why this PR

@GreenGrassBlueOcean after your #140 review found four defects our own verification had missed, we re-read every PR of ours against its neighbours rather than against its own claims. This is the remaining leftover on a merged PR.

• fix(connection): refuse WHO 4 dimension 22 in every send path #136 CHANGELOG, how it was found. Reading the sentence against send_to_gateway's body: it creates a session, calls _refuse_destructive_frame (which sets _last_error on that session and logs at ERROR) and returns None; the session is never returned, so "kept in last_error" is unobservable on that path. Text only; the behaviour is unchanged and already tested (test_send_to_gateway_refuses_without_connecting).
• Dropped after review (4f82518): the first revision also bounded _WHO4_WHERE_REGEX's #0#C to [1-4]. That was wrong: #0#C is the 4-zone central unit's configured zone selector, 1..99 (01..99), per MyHOME_Suite OPEN.db (#0#[ZA][ZB]), MHCatalogue 4695 (ZAZB_CENTRAL 1..99) and the oracle thermo-central-mode rows (*4*0*#0#2## → D1 02 …). The merged feat(heating): OWNHeatingCommand.actuators_status(where) builder for *#4*WHERE*20## #140 regex stays as it is.

Testing

• CHANGELOG.md only; scripts/sync_documentation.py --check in sync; ruff clean; tests/test_who4_actuators_status.py 60 passed (unchanged from master).

🤖 Generated with Claude Code

[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!

[GreenGrassBlueOcean]

Requesting changes. Checked against Legrand's WHO_4.pdf (p. 22), MyHOME_Suite OPEN.db, MHCatalogue.db (4695 / Object 191), BtExperience (BtObjects/thermalobjects.cpp:60-1015), and own-firmware-oracle (thermo-central-mode.tsv, thermo-zone-requests.tsv). The CHANGELOG.md correction on OWNCommandSession.send_to_gateway is right — connection = cls(gateway) is local and discarded on return None, so last_error is not observable on that path unless raise_on_error=True raises OWNProtocolError — and refusing #0#0 / #00 makes sense. Bounding #0#... to #0(?:#[1-4])?, however, rejects valid 4-zone central-unit addresses #0#5..#0#99 (#0#05..#0#99). CI 8/8 is green and the branch merges cleanly onto master (247f86e).

Blockers:

• #0#C on a 4-zone central unit is 1..99 (01..99), not 1..4. My #140 review wrote #0#1..#0#4 in passing (and OWNHeatingEvent.__init__ line 210 has an older comment calling #0#N "zone N of a 4-zone central unit"), which misled this bound. In the primary sources, #0#C is the 4-zone (4695) central unit's own configured zone selector, not a 1..4 sub-zone slot:
  • MyHOME_Suite OPEN.db (functional/open-db-coverage.md, ownkb:chunk:r000237) defines Four-zone control unit as Virtual #0#[ZA][ZB] and Advanced #0#[ZAZB].
  • MHCatalogue.db (OWN-DEV-0017 4695 and OWN-DEV-0018 Object 191, ownkb:chunk:r001952 / r001981) configures the unit's local zone with ZA = 0..9, ZB = 0..9 and ZAZB_CENTRAL in 1..99: a 4695 supervises up to four zones including its own local zone ZAZB, so a unit mounted as master of zone 5 or 12 is #0#5 (#0#05) or #0#12.
  • BtExperience (BtObjects/thermalobjects.cpp:60-1015, parseControlUnit4 / parseZone4, ownkb:chunk:moc-r001193) uses #0#C for central-unit commands and #Z#C for a zone controlled through 4-zone central unit C (#23#1 in libqtdevices TS10_1_0_23, cited in _zone_address_and_name).
  • own-firmware-oracle (results/F461/020011/checks/thermo-central-mode.tsv lines 17–31): *4*0*#0## emits $006D1000302C1F10000\r (D1 00 ...) and *4*0*#0#2## emits $006D1020302C1F10000\r (D1 02 ...) — a two-nibble ZAZB central address byte, not a 1..4 index.
• CHANGELOG.md line 17 grammar. Dropping send_to_gateway from the subject list left `OWNCommandSession.send` (and so `AsyncTcpTransport.send` and `keepalive`), `AsyncSerialTransport.send` refuses it — needs and AsyncSerialTransport.send refuse it.

Notes:

• Actuator Z#N vs central #Z#C on DIMENSION 20. WHO_4.pdf p. 22 specifies Z#N (Z = 0..99, N = 1..9), Z#0 and 0#0 (no leading # on actuator targets), and thermo-zone-requests.tsv + live F461 2.0.11 add *#4*Z*20## and *#4*#0*20##. Neither *#4*#0#C*20## nor *#4*#Z#C*20## has been replayed in the oracle yet, and in #Z#C the second token is the central selector C (1..99), not an actuator 0..9.
• 01 vs #01. Keeping two-digit virtual 01 while refusing #01..#09 splits the same [ZA][ZB] (ZA=0, ZB=1..9) convention; #00 and #0#0 / #0#00 are the only zero forms that name no valid zone or central unit.

Either dropping the heating.py / regex hunk (keeping #140 as-is and landing only the CHANGELOG.md send_to_gateway fix with the and ... refuse tweak) or bounding C in #0#C / #Z#C to 1..99 (01..99, rejecting #00 and #0#0 / #0#00) works for me.

[gdluck]

Thanks, you are right on the selector: #0#C is the 4-zone unit's configured zone (ZAZB, 1..99), not a 1..4 slot, and the D1 02 … byte in thermo-central-mode settles it. Took option one in 4f82518: heating.py and its tests are back to master (the merged #140 regex stays), only the CHANGELOG correction remains, with the subject list now "OWNCommandSession.send (and so AsyncTcpTransport.send and keepalive), and AsyncSerialTransport.send refuse it". Title and body updated accordingly.

The same misreading is in #137 (set_central_mode bounded #0#C to 1..4 and #Z#C to 1..9); I am correcting it there to the 1..99 / 01..99 selector, rejecting only #00, #0#0 and #0#00, and will note it on that PR.

[GreenGrassBlueOcean]

Approving. Both points from the review are addressed in 4f82518: heating.py and its tests are back to master (#140 untouched), and CHANGELOG.md accurately describes send_to_gateway's Dimension 22 refusal with the updated subject list. Good catch spotting the same #0#C / #Z#C selector bound in #137 as well. CI 8/8 green.