* Use forward entry setup with light platform
* Move sensor to forward entry setup
* Use forward entry setup with binary sensors
* Use forward entry setup with scene platform
* Remove import of unused functionality
* Move deconz setup in to setup entry
Create initial negative tests for setup entry
* Fix hound comment
* Improved tests
* Add test for scene platform
* Add test for binary sensor platform
* Add test for light platform
* Add test for light platform
* Add test for sensor platform
* Fix hound comment
* More asserts on sensor types
* Handle HomeKit configuration failure more cleanly
Add support for handling cases where HomeKit configuration fails, and give
the user more information about what to do.
* Don't consume the exception for a homekit.UnknownError
If we get an UnknownError then we should alert the user but also still
generate the backtrace so there's actually something for them to file in
a bug report.
When an OpenWrt device monitored via ubus is offline, this causes the
log to be flooded with several exceptions. Avoid this by catching
requests.exceptions.ConnectionError in addition to
requests.exceptions.Timeout.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
The Lokalise server has a bug that the internal portion of key
references was misinterpreted as a symfony key, and was getting auto
converted by the convert placeholders feature. Since we don't use this
we're turning it off to work around the bug.
* Initial working config entry with discovery
* No need for else
* Make sure that imported config doesnt exist as a config entry
* Improve checks to make sure there is only instance of deconz
* Fix tests and add new tests
* Follow upstream changes
Fix case when discovery started ongoing config entry and user completes setup from other path it was possible to complete discovered config entry as well
* Add test to make sure link doesn't bypass any check for only allowing one config entry
* Dont use len to determine an empty sequence
* Cleanup
* Allways get bridgeid to use as unique identifier for bridge
* Fix colorlog on windows
Modified the way logging is initialized to fix two things.
1. If the import of `colorlog` fails the logs will still be formatted
using the expected HASS log format.
2. Ensure that `logging.basicConfig` is called AFTER `colorlog` is
imported so that the default handler generated will be writing to the
wrapped stream generated when `colorama` is initialized. This allows
colored logging to work on Windows.
Added support for a `--log-no-color` command line switch in the event
that someone just wants to disable colored log output entirely.
* Fix line lengths
* Switch default value
Daniel Høyer Iversen
thelittlefireman
John Mihalic
Paulus Schoutsen
Kane610
Mark Coombes
Matthew Garrett
Otto Winter
Stijn Tintel
David Broadfoot
Ryan Bahm
Matt Schmitt
Jon Maddox
Aaron Bach
Johann Kellerman
Paulus Schoutsen
Rohan Kapoor
Sebastian Muszynski
ChristianKuehnel
Pascal Hahn
Viorel Stirbu
koolsb
Adam Mills
NovapaX
Michael Wei
Ben Randall
Nick Whyte
stephanerosi
Mister Wil
Thibault Cohen