1461617 - LWTConsumer should be consistent in fallback accentcolor (and/or line above tabs) when none/"#" is present in the lwtheme data

Status: NEW

(WebExtensions :: Themes, defect, P3)

Description

[:Gijs (he/him)]

STR:

1. install https://addons.mozilla.org/en-US/firefox/addon/dream-of-waves/ (comes with accentcolor: "#")
2. note black colour at the top of tabs
3. update the theme (accentcolor will be gone) and restart firefox or open a new window
4. note white colour at the top of tabs

AR:
colour changes

ER:
should be consistent


Historically this line has been black for these themes, but for the accentcolor a white fallback makes more sense. Not sure how best to proceed here.

Comment 1

[Tim Nguyen :ntim]

The easiest way to fix this is to use "white" or null as default on the AMO side. There's really no nice way to fix it on the Firefox front-end, as the front-end has no nice way to differentiate between an invalid color and black.

One possible hacky way to fix this on the Firefox side is:

  let div1 = doc.createElementNS(HTML_NS, "div");
  div1.style.color = "black";
  let span1 = doc.createElementNS(HTML_NS, "span");
  span1.style.color = cssColor;
  div1.appendChild(span1);

  let div2 = doc.createElementNS(HTML_NS, "div");
  div2.style.color = "white";
  let span2 = doc.createElementNS(HTML_NS, "span");
  span2.style.color = cssColor;
  div2.appendChild(span2);

  if (doc.defaultView.getComputedStyle(span1).color != doc.defaultView.getComputedStyle(span2).color) {
    return null;
  }
  return doc.defaultView.getComputedStyle(span1).color;


But since AMO has the information directly regarding the absence of accentcolor, it might be worth fixing this directly on the AMO side ?

Comment 2

[Dão Gottwald [:dao]]

From bug 1460287 comment 12: Options include using the text color instead of the accent color for the tab line, and/or letting authors set the tab line color explicitly.

Using the awfully poorly named "accent color" for the tab line is a hack, it wasn't originally intended to be used there.

Comment 3

[Mike de Boer [:mikedeboer]]

Marking as P3, but comment 2 seems to suggest it might be more serious than that and a potential P1. Feel free to mark it appropriately if that's the case!