6.4.1.14 20190203
* Add SetOption32 until SetOption49 diagnostic information to Status 3 report as replacement for second property value in SetOption property name
* Add Resolution property to Status 3 report providing previous SetOption second value property
* Add core version conditional compile options to provided PWM files (#4917)
* Add support for inverted buttons and inverted buttons without pullup (#4914)
6.4.1.11 20190124
* Remove command SetOption14 as it has been superseded by command Interlock
* Remove command SetOption63 as it has been superseded by command Interlock
* Add command Interlock 0 / 1 / 1,2 3,4 .. to control interlock ON/OFF and add up to 8 relays in 1 to 4 interlock groups (#5014)
6.4.1.8 20190107
* Change sonoff_template.h layout regarding optional module flags like ADC0
* Add command SetOption62 1 to force no Button/Switch pullup on dedicated modules. Currently only supported on Shelly2 (#4841)
6.4.1.2 20181228
* Change switch driver making it modular and introduce input filter (#4665, #4724)
* Add define DS18B20_INTERNAL_PULLUP to select internal input pullup when only one DS18B20 sensor is connected eliminating external resistor (#4738)
* Add variable %timestamp% to rules (#4749)
6.4.0.3 20181222
* Change Hass discovery to short MQTT messages as used by Hass 0.81 and up (#4711)
* Change FallbackTopic detection (#4706)
* Add define WIFI_SOFT_AP_CHANNEL in my_user_config.h to set Soft Access Point Channel number between 1 and 13 as used by Wifi Manager web GUI (#4673)
* Fix Alexa "this value is outside the range of the device". Needs power cycle and Alexa deletion/discovery cycle. (#3159, #4712)
6.3.0.17 20181211
* Add support for TheoV2 sensors as documented on https://sidweb.nl
* Add support for SDM220 (#3610)
* Enhance support for MPU6050 using DMP (#4581)
Fix the math in LoadAvg calculation. To keep average we take away one Nth of loops and add one Nth of current cycle ration to the average.
Now the math makes more sense... maximum load average on setoption36 0/1 = 999 (as it should be)
Remove the assumption of 100% loop duty cycle as we want an accurate loop load average and not a percentage as such - so if it overshoots the value stored for setoption36 this should be known by looking at the load average and not be distorted by reducing it to 100 basis points on the assumption that my_acitivity took as long as setoption36.
Theo Arends
dh.harald
lobocobra
Adrian Scillato
Erik
andrethomas2
andrethomas