Previous commit d4f7dfa successfully fixed the bug in which lights
would not turn off if a transition was specified, however if 'bri' is not
present in the payload of the PUT request set to deCONZ, then any
'transitiontime' ends up being ignored. This commit addresses the
unintended side effect by reintroducing 'bri', resulting in the following
payload:
{ "on": false, "bri": 0, "transitiontime": ... }
* Force to use access_token if hass.auth.active
* Not allow Basic auth with api_password if hass.auth.active
* Block websocket api_password auth when hass.auth.active
* Add legacy_api_password auth provider
* lint
* lint
* Update nut.py
Added input.frequency and a number of output parameters.
* Update nut.py
Fixed formatting issues
Added "devices" fields
* Separated "device.description" line to two lines.
* Update nut.py
Removed device.* sensors
When light.turn_off is invoked with a transition, the following payload was
sent to deCONZ via PUT to /light/N/state:
{ "bri": 0, "transitiontime": transition }
However, on recent versions of deCONZ (latest is 2.05.31 at the time of
writing) this does not turn off the light, just sets it to minimum brightness
level (brightness is clamped to minimum level the light supports without
turning it off).
This commit makes the code send this payload instead:
{ "on": false, "transitiontime": transition }
This works as intended and the light does transition to the 'off' state.
This change was tested with Philips Hue colored lights, IKEA colored lights
and IKEA white spectrum lights: they were all able to be turned off
successfully with the new payload, and none of them could be turned off with
the old payload.
* Use new trusted_proxies setting for X-Forwarded-For whitelist
* Only use the last IP in the header
Per Wikipedia (https://en.wikipedia.org/wiki/X-Forwarded-For#Format):
> The last IP address is always the IP address that connects to the last proxy,
> which means it is the most reliable source of information.
* Add two additional tests
* Ignore nonsense header values instead of failing
* Address inconsistent behavior between different controllers.
Correct issue with comparison that was preventing white value slider from being shown.
* Add white mode for Flux LED
* Call _bulb.turnOn() after bulb properties have been set to prevent immediate on action
* Only use existing brightness if rgb is None to prevent unexpected recalculation of passed rgb values.
* Remove blank line
* Undo change so current brightness is used in all cases.
