Neopixel White value is ignored when RGB set to 0

[EricGallimore]

If you want to make an RGBW pixel white by setting a white value like this:
neopixels.fill((0,0,0,255))

it will instead set the W value to 0. The offending logic is here:

        # If all components are the same and we have a white pixel then use it
        # instead of the individual components.
        if self._bpp == 4 and r == g == b:
            w = r
            r = 0
            g = 0
            b = 0

Perhaps this should change, such that the W pixel value is only inferred from RGB values if the W value is 0? The if statement could become
if self._bpp ==4 and r==g==b and w==0:

[ladyada]

we dont use RGBW pixels so much, but if you're up for trying and fixing we'd love a PR!

[zostay]

This is a really interesting problem. The suggested correction will fix the problem case in particular, but it will still cause problems when someone really wants to directly control all 32-bits of color. I think a better solution is:

1. Implement the change suggested by @EricGallimore

2. Add an additional check that disables the code all together in case someone wants direct control over the LEDs without any interference.

So the code becomes something like:

if self._bpp == 4 and self._casual == 1 and r==g==b and w==0:

(_casual is maybe not the right name, but I had to pick something...)

Then, if someone flips casual control bit to 0, this logic turns off and they can control all RGBW pixels without interference by the __setitem__ smart logic. That way if someone wants to burn the watts to turn their 8x8 RGBW pixel grid full on, they are welcome to do so. I hope they have a heat sink and a pair of solar glasses.

[caternuson]

Is this automagic behavior more of hindrance than a convenience? Another idea might be to get rid of it and just require explicitly setting what you want.

[ladyada]

thanx!

[caternuson]

@EricGallimore Thanks for pointing this out. Please try updating to the 1.5.2 version of the library:
https://github.com/adafruit/Adafruit_CircuitPython_seesaw/releases/tag/1.5.2