@tenphi/glaze 1.4.0 → 2.0.0

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/docs/api.md CHANGED
@@ -411,9 +411,9 @@ promotion for that color.
411
411
 
412
412
  ## Manual contrast level
413
413
 
414
- `contrastLevel` turns contrast from a two-tier switch into a `0–100` slider.
415
- It is a config field, so it works globally, per theme, per token, and through
416
- `extend()`:
414
+ `contrastLevel` puts the normal end of the two-tier contrast model on a `0–100`
415
+ slider. It is a config field, so it works globally, per theme, per token, and
416
+ through `extend()`:
417
417
 
418
418
  ```ts
419
419
  glaze.configure({ contrastLevel: 60 });
@@ -421,12 +421,17 @@ const theme = glaze(280, 80, { contrastLevel: 60 });
421
421
  glaze.color('#26fcb2', { contrastLevel: 60 });
422
422
  ```
423
423
 
424
- | Value | Meaning |
425
- | ---------------- | -------------------------------------------------------------------- |
426
- | `'auto'` | Default. The two-tier model: normal variants plus a high-contrast tier. |
427
- | `0` | Normal contrast, with **no** high-contrast tier. |
428
- | `100` | The high-contrast scheme as the only scheme. |
429
- | anything between | Resolved *at* that level. |
424
+ The level moves the **normal** variants — `light` and `dark` — along that slider.
425
+ It says nothing about high contrast: `lightContrast` / `darkContrast` stay the
426
+ true high-contrast resolution at every level, and
427
+ [`modes.highContrast`](#output-modes) alone decides whether they are emitted.
428
+
429
+ | Value | Meaning |
430
+ | ---------------- | -------------------------------------------------------------------------- |
431
+ | `'auto'` | Default. The normal variants take the authored normal entries as-is. |
432
+ | `0` | The same thing, reached through the slider — output-identical to `'auto'`. |
433
+ | `100` | The normal variants *are* the high-contrast ones, so the tier is dropped. |
434
+ | anything between | The normal variants are resolved *at* that level. |
430
435
 
431
436
  Levels `0` and `100` reproduce the classic `light` / `dark` and
432
437
  `lightContrast` / `darkContrast` output **bit for bit**.
@@ -446,6 +451,9 @@ The interpolated values are then fed through the ordinary resolve, so a
446
451
  contrast floor at any level is **solved**, not approximated, and `autoFlip`
447
452
  decides once per level.
448
453
 
454
+ These feed the two **normal** passes only. The high-contrast passes bypass the
455
+ level entirely, so the tier resolves exactly as it does in `'auto'` mode.
456
+
449
457
  Everything that does not vary by high contrast today — hue, saturation,
450
458
  `darkDesaturation`, `opacity`, `pastel` — is unaffected by the level.
451
459
 
@@ -495,24 +503,35 @@ directly.
495
503
 
496
504
  ### High-contrast output
497
505
 
498
- A manual level already carries the contrast preference, so there is no second
499
- tier to emit:
500
-
501
- - `resolve()` still returns four variants, but `lightContrast` / `darkContrast`
502
- **mirror** `light` / `dark`.
503
- - A **global** level turns high-contrast output off outright, so
504
- `tokens()` / `tasty()` / `json()` / `dtcg()` / `dtcgResolver()` / `tailwind()`
505
- emit no high-contrast tier. `modes.highContrast` goes inert — it reads as
506
- "emit a separate high-contrast set *when* contrast is automatic" — so leaving
507
- `highContrast: true` set in a build config while a user switches their
508
- preference from auto to manual is fine, and silent.
506
+ The level and the high-contrast tier are independent, and they compose: the
507
+ slider raises the baseline, and a `prefers-contrast: more` block still escalates
508
+ on top of whatever baseline the user has chosen.
509
+
510
+ - `resolve()` returns four variants at every level. `lightContrast` /
511
+ `darkContrast` are the **true** high-contrast resolution — bit-identical to
512
+ what `'auto'` resolves for those slots, at any level.
513
+ - [`modes.highContrast`](#output-modes) alone decides whether the tier is
514
+ emitted, exactly as in `'auto'` mode. A level does not turn it off.
509
515
  - `css()` always returns four strings and ignores `modes`, so its
510
- `lightContrast` / `darkContrast` strings repeat the normal declarations —
511
- an existing `@media (prefers-contrast: more)` block keeps matching the base
512
- block as the level changes, with nothing to rewire.
516
+ `lightContrast` / `darkContrast` strings now carry genuinely escalated values
517
+ at a mid level, where they previously repeated the normal declarations. An
518
+ existing `@media (prefers-contrast: more)` block starts doing real work.
513
519
  - A level on a **single theme or token** does not change which modes are
514
- emitted. Sibling themes in a palette keep their real high-contrast tier, and
515
- the manual one reports its own resolved values there.
520
+ emitted — `modes` is global-only, and a palette must not have one sibling
521
+ collapse the shared token structure.
522
+
523
+ | `contrastLevel` | `modes.highContrast: false` (default) | `modes.highContrast: true` |
524
+ | ----------------- | ------------------------------------- | ---------------------------------------- |
525
+ | `'auto'` / `0` | `light`, `dark` | + `lightContrast`, `darkContrast` |
526
+ | `1`–`99` | `light`, `dark` at the level | + the tier, unchanged by the level |
527
+ | `100` (global) | `light`, `dark` = the HC values | same — the tier is dropped as a duplicate |
528
+
529
+ That last row is the one exception. At level 100 the normal variants *are* the
530
+ high-contrast ones, so a separate tier would be an exact duplicate: a global
531
+ level of 100 emits a single light/dark set, and even an explicit
532
+ `modes: { highContrast: true }` on the call is suppressed, because it has nothing
533
+ different left to ask for. A level of 100 on a single theme or token still
534
+ reports into the tier, with values equal to its own normal ones.
516
535
 
517
536
  ### Driving it at runtime
518
537
 
@@ -526,8 +545,9 @@ slider.oninput = () => {
526
545
  };
527
546
  ```
528
547
 
529
- Because a manual level skips the two high-contrast passes, a manual resolve is
530
- also half the work of an `'auto'` one.
548
+ A manual resolve runs the same four passes as an `'auto'` one. Only level 100
549
+ skips the two high-contrast passes, because the normal ones already produced
550
+ those values.
531
551
 
532
552
  ### Clearing it, exporting it, base links
533
553
 
@@ -540,7 +560,9 @@ also half the work of an `'auto'` one.
540
560
  matches how `modes` and `states` are treated.
541
561
  - **Set the level globally, or on both sides of a base link.** A per-token level
542
562
  on a dependent but not on its base anchors the two at different levels — the
543
- same caveat as a per-instance `lightTone` override.
563
+ same caveat as a per-instance `lightTone` override. It is visible in the
564
+ high-contrast tier too, since a dependent's high-contrast variant reads its
565
+ base's high-contrast slots (and at level 100, the base's mirrored normal ones).
544
566
 
545
567
  `resolveContrastForLevel(spec, level, polarity?)` is exported for advanced use;
546
568
  see [Contrast solver](#contrast-solver).
@@ -1767,11 +1789,11 @@ boundaries, not the tone transfer.
1767
1789
  | `states.dark` | `'@media(prefers-color-scheme: dark)'` | State alias for dark mode tokens ([Tasty](https://tasty.style) export). Defaults to a media query so tokens react to the OS preference without registering custom states. |
1768
1790
  | `states.highContrast` | `'@media(prefers-contrast: more)'` | State alias for HC tokens ([Tasty](https://tasty.style) export). |
1769
1791
  | `modes.dark` | `true` | Include dark variants in exports. |
1770
- | `modes.highContrast` | `false` | Include HC variants. Inert while a global [`contrastLevel`](#manual-contrast-level) is set — a manual level has no separate HC tier, so this reads as "emit HC variants when contrast is automatic". |
1792
+ | `modes.highContrast` | `false` | Include HC variants. Independent of [`contrastLevel`](#manual-contrast-level) — the level positions the normal variants while these stay the true HC resolution. A global level of `100` is the exception: the tier would duplicate the normal set, so it is dropped. |
1771
1793
  | `shadowTuning` | `undefined` | Default tuning for all shadow colors. Per-color tuning merges field-by-field. |
1772
1794
  | `autoFlip` | `true` | Default for each color's `autoFlip`. When solving `contrast` (or applying a relative `tone` that overshoots `[0, 100]`), allow crossing to the opposite side instead of clamping. With `false`, only the requested direction is considered; unmet contrasts pin the tone to that direction's extreme (and emit a warning) and overshooting offsets clamp to the boundary. Override per color via [`autoFlip`](#autoflip). |
1773
1795
  | `inferRole` | `true` | Infer each color's [`role`](#roles) from its name when no explicit `role` is set. Set to `false` to opt out of name-based inference (the base-opposite and foreground-default fallbacks still apply). |
1774
- | `contrastLevel` | `'auto'` | Manual contrast level, `0`–`100`, replacing the two-tier high-contrast model with a slider. `0` reproduces the normal output and `100` the high-contrast output, bit for bit. See [Manual contrast level](#manual-contrast-level). |
1796
+ | `contrastLevel` | `'auto'` | Manual contrast level, `0`–`100`, putting the **normal** variants on a slider between themselves and the high-contrast ones. `0` reproduces the normal output and `100` the high-contrast output, bit for bit. The HC tier is unaffected. See [Manual contrast level](#manual-contrast-level). |
1775
1797
 
1776
1798
  | Method | Description |
1777
1799
  | ------------------------- | ----------------------------------------------------------------------------------- |
@@ -1779,7 +1801,7 @@ boundaries, not the tone transfer.
1779
1801
  | `glaze.getConfig()` | Snapshot the current resolved config (shallow copy). |
1780
1802
  | `glaze.resetConfig()` | Reset to defaults (also bumps the version counter). |
1781
1803
 
1782
- Themes and standalone color tokens keep a sparse local `GlazeConfigOverride` and merge the live global at resolve time for omitted fields. Authoring `.export(override?)` freezes the effective merge at call time; restored instances pin that freeze. `pastel` is instance-only (theme/token override or per-color) — not set via `configure()`. `contrastLevel` is the one field the freeze treats as a live preference: only an instance-authored level is written to the snapshot ([why](#two-rules-worth-knowing)).
1804
+ Themes and standalone color tokens keep a sparse local `GlazeConfigOverride` and merge the live global at resolve time for omitted fields. Authoring `.export(override?)` freezes the effective merge at call time; restored instances pin that freeze. `pastel` is instance-only (theme/token override or per-color) — not set via `configure()`. `contrastLevel` is the one field the freeze treats as a live preference: only an instance-authored level is written to the snapshot ([why](#clearing-it-exporting-it-base-links)).
1783
1805
 
1784
1806
  ---
1785
1807
 
@@ -1801,9 +1823,10 @@ palette.tokens({ modes: { dark: true, highContrast: true } });
1801
1823
 
1802
1824
  Resolution priority (highest first):
1803
1825
 
1804
- 1. A global [`contrastLevel`](#manual-contrast-level) — pins
1805
- `highContrast: false`, since a manual level leaves no separate high-contrast
1806
- tier to emit. `highContrast` is inert while it is set.
1826
+ 1. A global [`contrastLevel`](#manual-contrast-level) of `100` — pins
1827
+ `highContrast: false`, since the normal variants already *are* the
1828
+ high-contrast ones and the tier would duplicate them. Any other level leaves
1829
+ `highContrast` alone.
1807
1830
  2. Per-call `modes` option on `tokens` / `tasty` / `json`.
1808
1831
  3. `glaze.configure({ modes })` — global config.
1809
1832
  4. Built-in default: `{ dark: true, highContrast: false }`.
@@ -1873,7 +1896,7 @@ import {
1873
1896
  | Function | Description |
1874
1897
  | ------------------------------------- | ------------------------------------------------------------------------ |
1875
1898
  | `okhslToLinearSrgb(h, s, l)` | OKHSL (h: 0–360, s/l: 0–1) → linear sRGB tuple. |
1876
- | `okhslToSrgb(h, s, l)` | OKHSL → gamma-encoded sRGB tuple (0–1 per channel). |
1899
+ | `okhslToSrgb(h, s, l)` | OKHSL (h: 0–360, s/l: 0–1) → gamma-encoded sRGB tuple (0–1 per channel). |
1877
1900
  | `okhslToOklab([h, s, l])` | OKHSL → OKLab `[L, a, b]`. |
1878
1901
  | `oklabToOkhsl([L, a, b])` | OKLab → OKHSL. |
1879
1902
  | `srgbToOkhsl([r, g, b])` | Gamma sRGB (0–1) → OKHSL. |
@@ -1886,15 +1909,44 @@ import {
1886
1909
 
1887
1910
  ### Format writers
1888
1911
 
1912
+ Every writer takes `h` on 0–360 and `s` / `l` / `t` on **0–1** — the scale every
1913
+ conversion above *returns*, and the scale `resolve()` stores in a
1914
+ `ResolvedColorVariant`. The percentages are an output detail: the writers scale
1915
+ by 100 themselves where the CSS syntax asks for one. (The one thing on a
1916
+ different scale is the pair of tone transfers below, `toTone` / `fromTone`,
1917
+ which speak the 0–100 tone axis the authoring API takes.)
1918
+
1919
+ ```ts
1920
+ import {
1921
+ formatOkhsl,
1922
+ formatOkhst,
1923
+ formatRgb,
1924
+ formatHsl,
1925
+ formatOklch,
1926
+ } from '@tenphi/glaze';
1927
+
1928
+ formatOkhsl(280, 0.6, 0.95); // 'okhsl(280 60% 95%)'
1929
+ formatOkhst(280, 0.6, 0.95); // 'okhst(280 60% 95%)'
1930
+ formatRgb(280, 0.6, 0.95); // 'rgb(238.45 239.95 251.1)'
1931
+ formatHsl(280, 0.6, 0.95); // 'hsl(232.92 61.87% 95.99%)'
1932
+ formatOklch(280, 0.6, 0.95); // 'oklch(0.9571 0.015 280)'
1933
+ ```
1934
+
1935
+ So a producer composes with a writer directly, with nothing to rescale in
1936
+ between:
1937
+
1889
1938
  ```ts
1890
- import { formatOkhsl, formatRgb, formatHsl, formatOklch } from '@tenphi/glaze';
1939
+ const v = glaze.color('#7A4DBF').resolve().light;
1940
+ const { h, s, l } = variantToOkhsl(v);
1891
1941
 
1892
- formatOkhsl(280, 60, 95); // 'okhsl(280 60% 95%)'
1893
- formatRgb(280, 60, 95); // 'rgb(244 240 250)'
1894
- formatHsl(280, 60, 95); // 'hsl(280 60% 95%)'
1895
- formatOklch(280, 60, 95); // 'oklch(0.95 ... 280)'
1942
+ formatOkhst(v.h, v.s, v.t); // 'okhst(298.52 70.41% 45.02%)'
1943
+ formatOkhsl(h, s, l); // 'okhsl(298.52 70.41% 45.27%)'
1944
+ formatRgb(h, s, l); // 'rgb(122 77 191)'
1896
1945
  ```
1897
1946
 
1947
+ A value above 1 can only be pre-2.0 percentage-scale input, so the writers
1948
+ `console.warn` about it once per writer rather than emit a wrong color quietly.
1949
+
1898
1950
  To attach an alpha component, use `glaze.format(variant, format)` on a `ResolvedColorVariant` (which carries the `alpha` channel) instead of these raw writers.
1899
1951
 
1900
1952
  ### OKHST tone utilities
@@ -1914,12 +1966,12 @@ import {
1914
1966
 
1915
1967
  | Function | Description |
1916
1968
  | ------------------------------------------- | ------------------------------------------------------------------------- |
1917
- | `toTone(l, eps?)` | OKHSL lightness (0–1) → tone (0–100). Defaults to `REF_EPS`. |
1918
- | `fromTone(t, eps?)` | Tone (0–100) → OKHSL lightness (0–1). Inverse of `toTone`. |
1969
+ | `toTone(l, eps?)` | OKHSL lightness (0–1) → tone (**0–100**, the authoring scale — divide by 100 for `formatOkhst`). Defaults to `REF_EPS`. |
1970
+ | `fromTone(t, eps?)` | Tone (**0–100**) → OKHSL lightness (0–1). Inverse of `toTone`. |
1919
1971
  | `toneFromY(y, eps?)` / `yFromTone(t, eps?)` | Same transfer in luminance space (0–1). |
1920
1972
  | `okhstToOkhsl({ h, s, t })` | OKHST → OKHSL (`{ h, s, l }`). |
1921
1973
  | `okhslToOkhst({ h, s, l })` | OKHSL → OKHST (`{ h, s, t }`). |
1922
- | `variantToOkhsl(variant)` | `ResolvedColorVariant` (stores `t`) → `{ h, s, l, alpha }` for rendering. |
1974
+ | `variantToOkhsl(variant)` | `ResolvedColorVariant` (stores `t`) → `{ h, s, l }` for rendering (`alpha` stays on the variant). |
1923
1975
  | `REF_EPS` | Reference epsilon (`0.05`) for the canonical tone axis. |
1924
1976
 
1925
1977
  `ResolvedColorVariant` stores `{ h, s, t, alpha }` (tone, not lightness). Use
@@ -1964,6 +2016,6 @@ Exported constants: `APCA_PRESETS`, `APCA_HC_ENHANCEMENT` (`15`, the Enhanced Le
1964
2016
  | `maxIterations` | `18` | Max binary-search iterations per branch. |
1965
2017
  | `initialDirection` | higher-contrast side | Direction to search first (`'lighter'` or `'darker'`). |
1966
2018
  | `flip` | `false` | When `true`, try the opposite direction if the initial one doesn't meet the target. When `false`, only the initial direction is searched — unmet contrasts pin the result to that direction's extreme. |
1967
- | `preferInitial` | `false` | With `flip` on and **both** directions meeting the target, keep `initialDirection` instead of taking whichever result lands nearer `preferredTone`. Makes the chosen side independent of the target — what [`contrastLevel`](#manual-contrast-level) uses to keep a color on one side of its base across the ramp. The flip fallback is unaffected. |
2019
+ | `preferInitial` | `false` | With `flip` on and **both** directions meeting the target, keep `initialDirection` instead of taking whichever result lands nearer `preferredTone`. Makes the chosen side independent of the target — what [`contrastLevel`](#manual-contrast-level) uses to keep a color on one side of its base across the ramp (on the normal passes only; the high-contrast ones never probe). The flip fallback is unaffected. |
1968
2020
 
1969
2021
  Result: `{ tone, contrast, met, branch: 'lighter' | 'darker' | 'preferred', flipped? }`. `flipped: true` indicates the initial direction failed and the opposite direction satisfied the target.
@@ -642,24 +642,26 @@ In HC variants, Glaze bypasses the normal tone window and uses the full
642
642
  `[0, 100]` range. Edge tones can reach the edge; contrast floors have more room
643
643
  to solve.
644
644
 
645
- ### A slider instead of a tier
645
+ ### A slider through the tier
646
646
 
647
647
  The two-tier model assumes contrast is a binary OS preference. When contrast is
648
648
  instead a setting *your users* control, set
649
- [`contrastLevel`](api.md#manual-contrast-level) to a number and the HC tier
650
- becomes a `0–100` ramp: level `0` is your normal palette, `100` is the HC
651
- palette you already authored, and everything between is resolved at that level.
649
+ [`contrastLevel`](api.md#manual-contrast-level) to a number and your **normal**
650
+ palette slides along a `0–100` ramp: level `0` is the palette you authored, `100`
651
+ is the HC palette you authored, and everything between is resolved at that level.
652
652
 
653
653
  The authoring work is the same — the HC pairs above are what the slider ramps
654
- toward, so a palette tuned for high contrast needs no new fields. Two
655
- consequences worth planning for:
656
-
657
- - The separate HC tier stops being emitted, so wire the level itself into your
658
- build or runtime rather than a `prefers-contrast: more` block. Existing
659
- media-query wiring keeps working; it just receives the same values as the base
660
- block.
661
- - `contrastLevel: 0` is a useful state in its own right: normal contrast with no
662
- HC tier at all, for products that ship the slider but default it off.
654
+ toward, so a palette tuned for high contrast needs no new fields. Two things
655
+ worth planning for:
656
+
657
+ - The slider and the tier compose. The HC tier keeps resolving at full contrast
658
+ regardless of the level, so wire the level into your build or runtime *and*
659
+ keep your `prefers-contrast: more` block: a user who has raised the slider and
660
+ asked their OS for more contrast still gets the escalation on top. The one
661
+ exception is level `100`, where the two coincide and the tier is dropped as a
662
+ duplicate.
663
+ - `contrastLevel: 0` is the natural rest position for a slider you ship but
664
+ default off — it reproduces the `'auto'` output bit for bit.
663
665
 
664
666
  ## Checklist
665
667
 
package/docs/migration.md CHANGED
@@ -12,6 +12,7 @@ If you're starting from scratch, see [methodology.md](methodology.md) first —
12
12
  - [Wiring exports into the app](#wiring-exports-into-the-app)
13
13
  - [Prefix map strategies](#prefix-map-strategies)
14
14
  - [Migrating an existing color system](#migrating-from-an-existing-color-system)
15
+ - [Upgrading Glaze](#upgrading-glaze)
15
16
  - [Common pitfalls](#common-pitfalls)
16
17
 
17
18
  ## Choosing an export
@@ -296,6 +297,40 @@ Glaze gives you light/dark/HC for free, but only the light mode is matched again
296
297
 
297
298
  After migration, mark every default-only token (borders, shadows, disabled chip, code highlighting, etc.) `inherit: false`. Colored sibling themes only need the accent + tinted-surface chain — flagging the rest cuts the emitted token set per theme dramatically.
298
299
 
300
+ ## Upgrading Glaze
301
+
302
+ ### 2.0 — `format*` takes 0–1
303
+
304
+ The `format*` writers (`formatOkhsl`, `formatOkhst`, `formatRgb`, `formatHsl`,
305
+ `formatOklch`) used to take `s` / `l` / `t` on a 0–100 percentage scale while
306
+ everything that *produces* those values — `resolve()`, `variantToOkhsl`,
307
+ `srgbToOkhsl`, `oklabToOkhsl`, `okhslToSrgb` — returns them on 0–1. Composing
308
+ the two, which is the obvious thing to do, was off by 100x and failed silently:
309
+ `0.7` is a legal percentage, so you got a valid CSS string naming the wrong
310
+ (near-black) color.
311
+
312
+ The writers now take the 0–1 factors, so the whole library speaks one scale.
313
+ Drop the `* 100`:
314
+
315
+ ```diff
316
+ - formatOkhsl(v.h, v.s * 100, l * 100);
317
+ + formatOkhsl(v.h, v.s, l);
318
+
319
+ - formatOkhst(v.h, v.s * 100, v.t * 100);
320
+ + formatOkhst(v.h, v.s, v.t);
321
+ ```
322
+
323
+ Nothing else moves: `h` was always 0–360, `alpha` was always 0–1, and every
324
+ export method (`css()` / `tokens()` / `json()` / `tasty()` / `dtcg()` /
325
+ `tailwind()` / `glaze.format()`) emits the same colors — they were compensating
326
+ internally. Dropping the redundant `×100 ÷100` round-trip shifts float noise by
327
+ an ULP, which shows up nowhere except the (meaningless) hue term of a
328
+ fully-desaturated `hsl()` string.
329
+
330
+ A leftover `* 100` is not silent any more: a value above 1 cannot be a factor,
331
+ so the writers `console.warn` once per writer and you get an obviously-broken
332
+ `7040.68%` instead of a plausible wrong color.
333
+
299
334
  ## Common pitfalls
300
335
 
301
336
  | Symptom | Cause | Fix |
@@ -307,6 +342,7 @@ After migration, mark every default-only token (borders, shadows, disabled chip,
307
342
  | A relative `tone` like `'+48'` lands on the _wrong_ (darker) side of its base. | Overshooting offsets now mirror to the other side of the base by default (`autoFlip` inherits `autoFlip`). | Set `autoFlip: false` on the color to clamp to the boundary instead, or use `tone: 'max'`/`'min'` to force the extreme. |
308
343
  | `palette.tokens()` emits unexpected unprefixed names. | A `primary` was set on the palette (or per-call) and is duplicating the theme's tokens without prefix. | Pass `primary: false` to disable for that export, or rename `glaze.palette(themes, { primary })`. |
309
344
  | `console.warn: token "foo" collides with theme "bar"`. | Two themes resolved to the same output key under your prefix config. | Adjust the prefix map so each token is unique, or accept the first-write-wins behavior. |
345
+ | `console.warn: formatOkhsl() got a value above 1`. | Pre-2.0 percentage-scale input to a `format*` writer (a leftover `* 100`). | Pass the 0–1 factors the converters return — see [2.0 — `format*` takes 0–1](#20--format-takes-01). |
310
346
  | `console.warn: color "X" cannot meet contrast`. | The requested contrast target is physically unreachable for the color's hue/saturation against its base. | Lower the floor, change the base, or accept the closest passing variant. Use the `name` override on standalone colors to make the warning identifiable. |
311
347
 
312
348
  ## See also
package/docs/okhst.md CHANGED
@@ -98,6 +98,10 @@ hi(f) = hi + (100 - hi) * f
98
98
  `[15, 95] → [0, 100]`. The render `eps` is not interpolated, and a `false` window
99
99
  is already the full range and therefore level-invariant.
100
100
 
101
+ This continuous form applies to the normal variants only. High-contrast variants
102
+ are already at the full range, so the level does not widen them further — the
103
+ bypass above is what makes the high-contrast tier level-independent.
104
+
101
105
  ## Scheme adaptation
102
106
 
103
107
  Each regular color has an adaptation `mode`:
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@tenphi/glaze",
3
- "version": "1.4.0",
3
+ "version": "2.0.0",
4
4
  "description": "OKHST-based color theme generator with WCAG contrast solving for light, dark, and high-contrast schemes",
5
5
  "type": "module",
6
6
  "main": "./dist/index.cjs",