0.114: Dark mode, Open Z-Wave progress and more automation & scripts

(@bdraco – #37949) (recorder docs)

A sync will be required for source selection to work after this update. The
keyword to select source will also have changed from “Set mode on TV to..”
to “Set input on TV to..”

(@elupus – #35753) (google_assistant docs)

Due to the new 2FA endpoint, the YAML configuration is no longer supported.

In addition, due to some API changes, if you are upgrading from a previous
version, you will need to navigate to the integration page at startup and
re-configure Blink.

Your username and password will be imported, but you need to submit a 2FA token
(emailed to you by Blink) in order to continue setting up the integration.

(@fronzbot – #38027) (blink docs)

TensorFlow has been upgraded to TensorFlow 2.

TensorFlow 1 object detection models are not compatible with TensorFlow 2,
therefore you’ll will need to walk through the setup documentation again and
choose a new model from the Model Zoo.

(@hunterjm – #38384) (tensorflow docs)

A switch has been added to the Pi-hole integration. Because of the switch being
added, the pi_hole.enable service is removed.

The pi_hole.disable service remains in order to support disabling the service
for a given amount of time, but now it relies on the switch entity id instead
of the name set by the user, so the name field is removed.

With this change, a switch entity switch.pi_hole will be added if you
set up the integration with an API key and default name (Pi-hole).

Now, you can use switch.turn_on and switch.turn_off to enable and disable
the Pi-hole service permanently. To disable the service for a given time,
use pi_hole.disable with entity_id: switch.pi_hole and duration: 00:00:15.
To target all Pi-hole services, use entity_id: all.

(@shenxn – #35605) (pi_hole docs)

The Volumio integration is now configured through the UI. If you previously had
it in configuration.yaml, please remove it and configure through the
Integrations panel.

(@OnFreund – #38252) (discovery docs) (volumio docs)

The hvac_action() property has been removed from Evohome zones as this data
was inaccurate – the algorithm used to provide this data was not
reliable, and there is no reasonable hope for a satisfactory alternative.

You will have no indication if a zone is actively calling for heat
(CURRENT_HVAC_*), nor if the heat source is currently active. However,
you will continue to have the hvac_mode() property, which will indicate if the
heating is enabled, or not (HVAC_MODE_*).

This will be most obvious by the absence of this information in the UI.

There is no intervention required for this breaking change.

(@zxdavb – #38244) (evohome docs)

The unique id is being updated for existing sensors. You will need to clean up
the stale entity from the entity registry via the GUI.

(@ChrisMandich – #38138) (flume docs)

The meteo_france integration moved from a web scraping architecture to a
API-based architecture. With this major refactor the following change could
break:

• The thunder chance entity is no longer available (no longer in the API results).
• The next rain entity state is now a datetime instead of a delay.
• next rain attributes have been simplified.
• weather_alert: only one entity created by department. Was by city in the
  previous version and could create duplicates.
• weather alert attributes have been simplified.
• weather platform is showing the current day forecast in the first position of
  daily forecast (instead of tomorrow forecast).

(@oncleben31 – #37737) (meteo_france docs)

When you have multiple Bond hubs, each with its own set of devices, those
devices have IDs that the Bond integration uses to generate unique entity IDs
in Home Assistant. Unfortunately, those IDs are only unique within the hub.
So it is possible that unique IDs generated for entities from different hubs
will not be unique and collide.

This has been fixed by prefixing device ID with a hub ID and the resulting ID
is now truly unique. After restarting you will see each bond entity twice.
One which is live with the new unique ID. The second which is disabled with old
unique ID.

How to make it work:

1. Remove the disabled entity (note its friendly ID, e.g., cover.shades)
2. Optionally rename the generated friendly ID on the live entity, so that your
   automations continue to work. For example, your live entity will have ID
   cover.shades_2. Rename it to cover.shades.

(@prystupa – #38496) (bond docs)

The hlk-sw16 entity ids will change upon update and the old entities will need
to be removed manually.

(@jameshilliard – #37190) (hlk_sw16 docs)

The Slack attachments framework was deprecated in 0.108.0 and is now officially
removed. The blocks framework should be used instead.

(@bachya – #38139) (slack docs)

Per ADR-0010, IQVIA can no longer be configured via YAML and must be configured
from the UI. Existing IQVIA users have already had their integrations imported
and only need to remove IQVIA-related items from configuration.yaml.

(@bachya – #38141) (iqvia docs)

Processes spawned from the shell_command integration (the command executes)
now timeout after 60 seconds instead of running forever.

(@bdraco – #38491) (shell_command docs)

The command_line cover, notify and switch platforms, now timeout after 15
seconds (by default) instead of blocking forever.

(@bdraco – #38497) (command_line docs)

If time abruptly moves forward or backward, time pattern listeners will only
be adjusted after they were previously scheduled to fire.

This avoids the need for every time pattern listener to check for this every
second.

(@bdraco – #38021)

Support for legacy logbook events created before 0.112 is removed.

(@bdraco – #37822) (logbook docs)

This change is actually not a breaking change, but it is making up for the fact
that a previous change was that was not marked as such.

It turns out there are valid use cases where an automation needs to be turned
off, so that it will not trigger, but where previous runs may still be active
that should be allowed to complete. For example, an automation that performs a
reasonably lengthy procedure that should not be aborted midstream, otherwise
it would leave the system in an undesirable state. Or a queued automation that
needs to handle all previous events, some of which may not be complete when
the automation is turned off.

To accommodate for these use cases, this release introduces adds an optional
parameter to the automation.turn_off service, namely stop_actions, which
defaults to true. It controls whether or not active runs should be stopped when
the automation is turned off.

(@pnbruckner – #38436)