Model 100 LED regression: only some LEDs light on boot

[iliana]

Describe the bug
In the default sketch, the first LED effect is LEDOff. On my firmware sketch, I enable a solid color LED effect immediately (and have no other LED effects).

Somewhere in becf816...c1acb3b (this comparison looks confusing due to the LEDEffect-DigitalRain repo merge) seems to introduce a regression for this firmware. On power up, some or none of the LEDs light. I can hit my LED key (bound to Key_LEDToggle because I have only one effect) twice, and all LEDs light as expected.

When some LEDs do light, there are patterns. Generally it is the right 2 columns of the right half, and only the LED key on the left half. Sometimes we also get the tab and left Fn key.

7e2a767 looks potentially related? I am not sure.

To Reproduce

1. Flash my sketch.

Expected behavior
I expected all the LEDs to light.

Environment (please complete the following information):

• OS: macOS
• Version: 12.6
• Device: Keyboardio Model 100

[iliana]

Adding this to my setup() function reliably works around this issue:

LEDControl.disable();
delay(100);
LEDControl.enable();

[obra]

@tlyu Do you think this might be related to some of your recent rework on startup?

[tlyu]

@tlyu Do you think this might be related to some of your recent rework on startup?

Yeah, it's possible. I'll try to reproduce, and I can try adding a poll for both ATtinys to start responding before exiting the Model100 setup (with a reasonable timeout in case someone is running with the right half disconnected).

[tlyu]

I can't reproduce it by setting my default LED mode to one of the solid colors on a default Model 100 sketch. I'm going to assume that the custom sketch speeds things up enough that the solid color mode writes the LEDs before the ATtinys have booted up.

[iliana]

That's my assumption as well. (Even with the same firmware build on my sketch, different numbers of LEDs come up on the left half on each boot, so I think the 100ms workaround I added is probably overkill.)