What's Changed
Integration
- A discovered openccu-loom daemon no longer re-points a configured entry by itself. When a
daemon announced a CCU serial that an entry already uses, the entry silently switched to that
daemon's address and reloaded — and sent its sign-in token there. A second daemon for the same CCU
(a test build, another installation) could take over a working setup that way, and so could any
host on the network that announced the serial. Now the entry stays as it is and a discovery card
asks whether the daemon moved; only a confirmation switches it. A daemon whose announced instance
id differs from the one the entry knows is recognised as another daemon and ignored
(needs openccu-loom with SukramJ/openccu-loom#1131; the id is learned at setup and on a confirmed
move) (#1386). - Sub devices are now Home Assistant child devices. A channel group and the schedule of a
device used to be separate devices linked withvia_device; they are now child devices of the
device they belong to, as HA 2026.9 introduced for parts of one product
(HA developer blog).
HA converts the existing entries in place, so everydevice_id, area, custom name and
automation survives. What changes: a child carries no manufacturer, model, firmware or serial of
its own, inherits the device's area unless the channel group has its own room, and is deleted or
disabled together with the device. Its name drops the device prefix, because HA composes the full
name from device and child itself: the schedule child is now calledSchedule/Zeitplan, an
unnamed channel groupGroup 2/Gruppe 2, a named group keeps its name. Thefriendly_name
state attribute of their entities therefore no longer starts with the device name. The link from a
device to the central stays avia_devicelink. - HmIP-WGT and HmIP-WGTC show their heating schedule. They also have a switch schedule, and the
climate entity and the climate schedule services only knew that one. The climate entity and its
services now use the heating schedule.get_scheduleandset_schedulestay on the switch schedule
of these devices. The new heating schedule sensor is called "Climate week profile". The config panel
still shows the switch schedule only. - The Messages card shows the CCU's text for service messages. It rendered the raw parameter
name from the CCU alarm, so a sabotage of an HM-Sec-SCo readERROR-7where the CCU shows
"Sabotage". The card now uses the same translated text as the alarm list and the CCU dashboard
(#1356) - The CCU backup location survives a reload of the integration. After a reload — changed
options, a reconfigure, a reauth — the backup location "CCU " was gone from the Home
Assistant backup settings until Home Assistant was restarted, and automatic backups no longer
included a CCU backup. The integration told Home Assistant about its backup agents while the
entry was still being set up, and Home Assistant only accepts agents of a loaded entry - openccu-loom (Beta): a CCU switched to the daemon behind an openccu-lite box starts again. The
switch kept the CCU user name and password in the entry, and the box connection refuses any
credential beside its box token, so setup failed withbox_token is the only credential behind the box. A backend switch no longer carries the previous backend's credentials along; existing
entries are cleaned up automatically, and switching back asks for the credentials again
(openccu-lite#20) - Changing an entity ID keeps the entity in place. Home Assistant removed every entity of the
integration on an entity ID change and added it again, so the entity dropped its subscriptions and
reloaded its value. Since HA 2026.11 an integration can take the change in place
(HA developer blog),
and all entities of the integration now do. - Repairs belong to their integration entry. Home Assistant deletes the repairs of an entry
together with the entry and hides them while the entry is disabled
(HA developer blog).
Repairs raised before this version are not linked yet; they are linked when they are raised again. switch_set_on_timeworks on a switch that is already on. On a generic switch (CUxD and Homematic
channels without a device-specific switch) the on time was written at once, and the following
switch.turn_onsent nothing because the switch was already on. The action now stages the on time on
every switch, andswitch.turn_onsends it together with the switch-on, also when the switch is already
on (#1378).- Several CCUs start reliably at Home Assistant startup. With more than one config entry, the entries set
up at the same time and could register the static paths of the cards or the configuration panel twice;
Home Assistant refused the second registration, so one entry failed to set up at random and only a restart
recovered it. The registration now runs once, one entry after the other
(#1376). - The climate schedule shows the heating times as periods. The simple week profile used the
temperature that runs longest on a day as base temperature, so a profile whose comfort phase is the
longer one showed the setback hours as coloured periods and the heating phase as grey background. The
card and the configuration panel now take the lowest temperature of the day as base; the schedule on the
device is unchanged
(frontend#129).
Dependencies
Bump aiohomematic to 2026.11.6
- Support for new devices: thermostats HmIP-WT, KW-STATH, KW-UKETRV, HmIPW-WGT and HmIPW-WGTC, dimmers
HmIP-WUA, ELV-SH-WUA and HmIP-UDI-PB2 - Fix: HmIP-DLD and HmIP-DLP report a jammed lock, HmIP-DLP exposes
ERROR_JAMMED - Fix: HmIP-WSM keeps its run time on current firmware, HmIP-BDT-I dimmer uses the correct channels
- Fix: the switch schedule of HmIP-WGT and HmIP-WGTC has the correct targets and accepts entries
- Fix: HmIPW-WTH and HmIPW-SCTHD use the limits configured on the device
- Fix: tilt-only changes turn the slats of a fully closed HmIPW-DRBL4 again, a regression of 2.12.2 (#1357)
- Fix: service messages of enum parameters carry the CCU's value text, so the hub sensor
service_messagesshowsSabotageinstead ofERROR-7(#1356) - Fix: switching the week program of a heating group (HM-CC-VG-1) no longer breaks its device settings page in the CCU WebUI (#1379)
- Fix: a timed switch-on reaches a generic switch that is already on (#1378)
Bump aiohomematic-config to 2026.11.1
- Fix: the climate schedule commands of the config panel (
get_climate_schedule,set_climate_schedule_weekday,set_climate_active_profile) reach the heating schedule of HmIP-WGT and HmIP-WGTC, which also have a switch schedule
Bump openccu-loom-client to 2026.11.10
- For the openccu-loom backend (Beta) only; no runtime effect on the direct-CCU backend, where the client is not loaded. Fixes
switch_set_on_time,light_set_on_timeandvalve_set_on_time, which had no effect on that backend, climate entities that were not created there with this release, and device actions, device triggers and the per-device central-link services, which failed there; HmIP-WGT and HmIP-WGTC get their heating schedule with openccu-loom 0.97.1 or newer - Fix: service messages of enum parameters carry the CCU's value text on the openccu-loom backend too, so the hub
sensorservice_messagesshowsSabotageinstead ofERROR-7(openccu-loom 0.97.2 or newer) - Fix: after the openccu-loom daemon restarts, the integration waits until the daemon has brought its devices up
before it reads them again. It created entities for hidden parameters (UNREACH,BOOTED, …) that a cold start
never creates, and logged one "does not generate unique IDs" error per entity - Wire bindings for openccu-loom 0.100.0, which replaced the built-in measurement history with measurement forwarding
- Fix: downloading the diagnostics failed with HTTP 500 on the openccu-loom backend as soon as the daemon had
recorded an incident (#1368) - Fix: the incident list of the integration panel and of the system-health card shows text, time and (with
openccu-loom REST API 16.4.0 or newer) type on the openccu-loom backend; the rows were blank but for their colour - Fix:
switch_set_on_timereaches a generic switch on the openccu-loom backend, which dropped the on time
(#1378) - Fix: lights with effects (effect, RGBW and signal lights) show their effect list on the openccu-loom backend, and
switching one on with an effect switches it on; the effect used to go out without a switch-on
(#1385) - Reads the instance id an openccu-loom daemon announces (openccu-loom 0.106.0 or newer), which the discovery of a
moved daemon uses to tell two daemons for the same CCU apart
(#1386)