@josueavalosjim/taste-check 0.5.1 → 0.5.2

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/README.md CHANGED
@@ -227,6 +227,15 @@ page's own scripts, which is where a theme belongs because the page reads it at
227
227
  boot. `after` runs once there is a document, for opening a panel or focusing a
228
228
  field. Each state's setup is removed before the next one, so they cannot leak.
229
229
 
230
+ **A foreground composites over its own background. An edge measures against
231
+ what is outside it.** Those are different questions. Text sits on the thing
232
+ behind it, so its own background belongs in the stack. A border is a boundary,
233
+ and 1.4.11 asks whether that boundary can be told apart from what is adjacent
234
+ to it, which is the page rather than the control's own fill. A solid button
235
+ that sets its border to its own background colour measures 1.00:1 the first
236
+ way and 14.68:1 the second, and only one of those is the edge anybody sees.
237
+ Border properties are handled this way for you.
238
+
230
239
  Use `againstParent` when the thing being measured is a fill rather than a
231
240
  foreground. A selected row with its own background measured against itself
232
241
  scores 1.00, which is a pass that means nothing.
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@josueavalosjim/taste-check",
3
- "version": "0.5.1",
3
+ "version": "0.5.2",
4
4
  "description": "Design review in CI with a line down the middle: measured checks that gate the build (WCAG contrast from your tokens or from a real rendered page, one-off values in your markup) and a fresh-eyes model judge whose verdicts stay advisory. Zero dependencies. Ships no design rules of its own.",
5
5
  "keywords": [
6
6
  "accessibility",
package/src/runtime.mjs CHANGED
@@ -17,6 +17,10 @@
17
17
  * 2. It can measure a state, because it can put the page into one first.
18
18
  * A theme, an opened panel, a focused input.
19
19
  *
20
+ * A foreground composites over its own background. An edge measures against
21
+ * what is outside it. Those are different questions and getting them the same
22
+ * way round produces confident nonsense in both directions.
23
+ *
20
24
  * A selector that matches nothing is a failure, not a skipped target. So is a
21
25
  * border with no width: you asked for the contrast of an edge that is not
22
26
  * being drawn, and the honest answer is that the question is wrong.
@@ -42,7 +46,18 @@ const MEASURE = `(targets) => targets.map(({ selector, prop, againstParent }) =>
42
46
  if (!width) return { error: 'has a ' + prop + ' but no width, so no edge is drawn' };
43
47
  }
44
48
  const layers = [];
45
- let node = againstParent ? el.parentElement : el;
49
+ // Two reasons to start one level up.
50
+ //
51
+ // A fill measured against itself scores 1.00, so againstParent hands us the
52
+ // parent instead.
53
+ //
54
+ // A border is the same, for a different reason. 1.4.11 asks whether the
55
+ // boundary of a control can be told apart from what is adjacent to it, and
56
+ // what is adjacent is the page, not the control's own fill. A solid button
57
+ // that sets its border to its own background colour measures 1.00:1 against
58
+ // that background: true, and an answer to a question nobody asked.
59
+ const outside = againstParent || prop.indexOf('border') === 0;
60
+ let node = outside ? el.parentElement : el;
46
61
  if (!node) return { error: 'has no parent to measure against' };
47
62
  while (node) {
48
63
  const bg = getComputedStyle(node).backgroundColor;