On Tue, May 23, 2023, at 1:08 AM, Jens-Ulrik Petersen wrote:
I actually would prefer that we color both, and make it obvious that
"root" is special. We should account for common color-blindness
issues, though.
Sure, I think I agree: perhaps purple for root?
I think if we avoid the need to distinguish between redish and greenish, it's OK. Redish includes red, pink, magenta. Greenish includes green, cyan, sky blue.
Ergo, you can use green or red, you just don't want to make the A vs B red and green because they will look the same to anyone with a red/green color discrimination limitation.
Much less common is blue/yellow. I don't have data handy but off the top of my head, tritonopia and tritonomaly cases will notice the difference between purple vs orange OK. We do have publicly available math to predict the discrimination of various nopias and nomalies; I'm not sure exactly how it's implemented in GIMP but there is a "color deficient vision" filter should give an adequate idea whether or not a design has expected/sufficient/desired differentiation of elements based on color. (We don't actually know what anyone's color experience is, as it turns out. That's what I mean by this.)
I am all for "color blind testing" (though I am not completely sure that "color-blind" is the right term here
though I am not an a11y expert - I thought color blind is more about differentiating different colors like green and red,
but if you mean visual impairment/contrast/readability then I completely agree).
It's common vernacular. But it's more interesting than this term because it suggests one variety or effect. There are more than several. If we consider "color blind" means total lack of color discrimination, or monochromat - they are rare. I don't even know the number. Dichromats are much more common, where these are broken down into whether it's the long, medium, or short wavelngth cone is missing (entirely). The world does have some color, we think, at least there's discrimination possible. But it's of course limited without a third color receptor. Quite a lot more common are tricromats with anomalous spectral sensitivity of one of the receptors, i.e. the long wavelength cone might be green shifted, so it's more sensitive to yellows than the "standard observer". This then questions the whole age old concept of the standard observer, and for a while now it's been suspected we need more than one standard observer - because, well, they did all these tests in the early 1900's with something like 50 people. Seriously. And that's the data still largely used today. Anyway...
I tend to call it a color discrimination variance. Or limitation.
Oh and there are such folks as quadchromats. They have four color receptors. And then still there's the entire non-human animal kingdom full of completely different receptor peak wavelenth sensitivities, including tetrachromats and pentachromats.
I think in the end it will come down also to wider user testing since there are so many different terminals
and color palettes around.
Anyway that's why I proposed green since it seems to have reasonable contrast for both light and dark terminals (unlike blue/cyan/yellow often).
I assume that may also be why Ubuntu and Nixos went with green.
Green is an efficient color choice. It tends to appear to the brightest. Part of this relates to the luminosity function of human vision which has a peak wavelength that happens to be the same as the medium wavelength photo receptor (i.e. green). So given the same amount of radiant energy emitted across the visible spectrum, green will appear to be the brightest.
Light purple is OK, Blue, indigo, or yellow tends to be harder to to detect complex shapes (like letters and numbers) but I'm not sure of the reason(s).