Autodark - revised version

[romanseidl]

This is now a heavily revised version - it replaces the old pull request.

• updated to the current master
• the subscription is as lightweight as possible. all the work is in a seperate timer that gets called every second
• the lifecycle issue with the light sensor subscription has been resolved by canceling the subscription for the setting screen in a PopScope instead of in dispose.

I hope this is now a good basis for review.

[micw]

I have read through the code. Besides some minor issues (e.g. leak of the listener in some cases), I'm not happy that this would break native display-off.
More testing would be needed but potentially it could be possible during night time:
a) to switch the display off but let the CPU awake to read the sensor
b) maybe also sleep the CPU for 15s or so and read the sensor in intervals. Not sure how reliable this is, especially how often alarms are triggered in sleep mode.

I'd say a) is good enough. A non-sleeping CPU would be kind of ok when the device is wall-powered. But keeping the display on means a light source through the night time which would be a bad thing.

[romanseidl]

This is now a heavily revised version - it replaces the old pull request.

• updated to the current master
• the subscription is as lightweight as possible. all the work is in a seperate timer that gets called every second
• the lifecycle issue with the light sensor subscription has been resolved by canceling the subscription for the setting screen in a PopScope instead of in dispose.

I hope this is now a good basis for review.

[micw]

I have read through the code. Besides some minor issues (e.g. leak of the listener in some cases), I'm not happy that this would break native display-off.
More testing would be needed but potentially it could be possible during night time:
a) to switch the display off but let the CPU awake to read the sensor
b) maybe also sleep the CPU for 15s or so and read the sensor in intervals. Not sure how reliable this is, especially how often alarms are triggered in sleep mode.

I'd say a) is good enough. A non-sleeping CPU would be kind of ok when the device is wall-powered. But keeping the display on means a light source through the night time which would be a bad thing.

[romanseidl]

This is now a heavily revised version - it replaces the old pull request.

• updated to the current master
• the subscription is as lightweight as possible. all the work is in a seperate timer that gets called every second
• the lifecycle issue with the light sensor subscription has been resolved by canceling the subscription for the setting screen in a PopScope instead of in dispose.

I hope this is now a good basis for review.

[micw]

I have read through the code. Besides some minor issues (e.g. leak of the listener in some cases), I'm not happy that this would break native display-off.
More testing would be needed but potentially it could be possible during night time:
a) to switch the display off but let the CPU awake to read the sensor
b) maybe also sleep the CPU for 15s or so and read the sensor in intervals. Not sure how reliable this is, especially how often alarms are triggered in sleep mode.

I'd say a) is good enough. A non-sleeping CPU would be kind of ok when the device is wall-powered. But keeping the display on means a light source through the night time which would be a bad thing.

[romanseidl]

This is now a heavily revised version - it replaces the old pull request.

• updated to the current master
• the subscription is as lightweight as possible. all the work is in a seperate timer that gets called every second
• the lifecycle issue with the light sensor subscription has been resolved by canceling the subscription for the setting screen in a PopScope instead of in dispose.

I hope this is now a good basis for review.

[micw]

I have read through the code. Besides some minor issues (e.g. leak of the listener in some cases), I'm not happy that this would break native display-off.
More testing would be needed but potentially it could be possible during night time:
a) to switch the display off but let the CPU awake to read the sensor
b) maybe also sleep the CPU for 15s or so and read the sensor in intervals. Not sure how reliable this is, especially how often alarms are triggered in sleep mode.

I'd say a) is good enough. A non-sleeping CPU would be kind of ok when the device is wall-powered. But keeping the display on means a light source through the night time which would be a bad thing.

[romanseidl]

This is now a heavily revised version - it replaces the old pull request.

• updated to the current master
• the subscription is as lightweight as possible. all the work is in a seperate timer that gets called every second
• the lifecycle issue with the light sensor subscription has been resolved by canceling the subscription for the setting screen in a PopScope instead of in dispose.

I hope this is now a good basis for review.

[micw]

I have read through the code. Besides some minor issues (e.g. leak of the listener in some cases), I'm not happy that this would break native display-off.
More testing would be needed but potentially it could be possible during night time:
a) to switch the display off but let the CPU awake to read the sensor
b) maybe also sleep the CPU for 15s or so and read the sensor in intervals. Not sure how reliable this is, especially how often alarms are triggered in sleep mode.

I'd say a) is good enough. A non-sleeping CPU would be kind of ok when the device is wall-powered. But keeping the display on means a light source through the night time which would be a bad thing.

[romanseidl]

This is now a heavily revised version - it replaces the old pull request.

• updated to the current master
• the subscription is as lightweight as possible. all the work is in a seperate timer that gets called every second
• the lifecycle issue with the light sensor subscription has been resolved by canceling the subscription for the setting screen in a PopScope instead of in dispose.

I hope this is now a good basis for review.

[micw]

I have read through the code. Besides some minor issues (e.g. leak of the listener in some cases), I'm not happy that this would break native display-off.
More testing would be needed but potentially it could be possible during night time:
a) to switch the display off but let the CPU awake to read the sensor
b) maybe also sleep the CPU for 15s or so and read the sensor in intervals. Not sure how reliable this is, especially how often alarms are triggered in sleep mode.

I'd say a) is good enough. A non-sleeping CPU would be kind of ok when the device is wall-powered. But keeping the display on means a light source through the night time which would be a bad thing.

[romanseidl]

This is now a heavily revised version - it replaces the old pull request.

• updated to the current master
• the subscription is as lightweight as possible. all the work is in a seperate timer that gets called every second
• the lifecycle issue with the light sensor subscription has been resolved by canceling the subscription for the setting screen in a PopScope instead of in dispose.

I hope this is now a good basis for review.

[micw]

I have read through the code. Besides some minor issues (e.g. leak of the listener in some cases), I'm not happy that this would break native display-off.
More testing would be needed but potentially it could be possible during night time:
a) to switch the display off but let the CPU awake to read the sensor
b) maybe also sleep the CPU for 15s or so and read the sensor in intervals. Not sure how reliable this is, especially how often alarms are triggered in sleep mode.

I'd say a) is good enough. A non-sleeping CPU would be kind of ok when the device is wall-powered. But keeping the display on means a light source through the night time which would be a bad thing.

[romanseidl]

This is now a heavily revised version - it replaces the old pull request.

• updated to the current master
• the subscription is as lightweight as possible. all the work is in a seperate timer that gets called every second
• the lifecycle issue with the light sensor subscription has been resolved by canceling the subscription for the setting screen in a PopScope instead of in dispose.

I hope this is now a good basis for review.

[micw]

I have read through the code. Besides some minor issues (e.g. leak of the listener in some cases), I'm not happy that this would break native display-off.
More testing would be needed but potentially it could be possible during night time:
a) to switch the display off but let the CPU awake to read the sensor
b) maybe also sleep the CPU for 15s or so and read the sensor in intervals. Not sure how reliable this is, especially how often alarms are triggered in sleep mode.

I'd say a) is good enough. A non-sleeping CPU would be kind of ok when the device is wall-powered. But keeping the display on means a light source through the night time which would be a bad thing.