* Add websocket-based non-polling variant for songpal
* linting fixes
* changes based on Martin's feedback
* Fix linting
* add backoff timer for reconnects, fix variable naming (I thought that this wouldn't matter for internals..)
* Remove poll configuration variable
* bump the version just to be sure, the previous release lacked a version file (required for setup.py)
* event foundation
* implement quirks
* lock zha-quirks version
* allow quirks handling to be toggled on and off
* revert event commit
* disable warning
* update requirements_all
* Remove fix in favor of #19141#19141 should be what ultimately corrects this issue.
* review comment
This works by adding `?check_same_thread=False` to the sqlite
connection string, as suggested by robbiet480@.
It upgrades pygtfs from homeassitant's forked 0.1.3 version to 0.1.5.
Fixes#15725
* radiotherm: bump version to 2.0.0
* radiotherm: change handling of transient errors from tstat
Radiotherm 2.0.0 now throws an exception when a transient error is
detected, instead of returning -1 for the field where the error was
detected. This change supports handling the exception.
This bumps to the new version of the waterfurnace API. In the new
version the unit id is no longer manually set by the user, instead it
is retrieved from the service after login. This is less error prone as
it turns out discovering the correct unit id is hard from an end user
perspective.
Breaking change on the config, as the unit parameter is removed from
config. However I believe the number of users is very low (possibly
only 2), so adaptation should be easy.
This bumps to the new version of the waterfurnace API. In the new
version the unit id is no longer manually set by the user, instead it
is retrieved from the service after login. This is less error prone as
it turns out discovering the correct unit id is hard from an end user
perspective.
Breaking change on the config, as the unit parameter is removed from
config. However I believe the number of users is very low (possibly
only 2), so adaptation should be easy.
After six months the chevy website finally has been reimplemented to
something that seems to work and is stable. The backend library has
been updated thanks to upstream help, and now is working again.
* Updated Yale Smart Alarm platform to use Yale's new API which replaces the deprecated version. Bumped yalesmartalarmclient to v0.1.5.
* Update requirements
* Initial hlk-sw16 relay switch support
* remove entity_id and validate relay id's
* Bump hlk-sw16 library version and cleanup component
* refactor hlk-sw16 switch platform loading
* Use voluptuous to coerce relay id to string
* remove force_update for SW16Switch
* Move to callback based hlk-sw16 relay state changes
* fix hlk-sw16 default port and cleanup some unused variables
* Refactor to allow registration of multiple HLK-SW16 device
* Store protocol in instance variable instead of class variable
* remove is_connected
* flake8 style fix
* Move reconnect logic into HLK-SW16 client library
* Cleanup and improve logging
* Load hlk-sw16 platform entities at same time per device
* scope SIGNAL_AVAILABILITY to device_id
* Fixes for connection resume
* move device_client out of switches loop
* Add timeout for commands and keep alive
* remove unused variables
