@mkbabb/glass-ui 4.0.1 → 4.2.0

package/dist/styles/tokens/scheme-motion.css

@@ -1,5 +1,10 @@
 /* tokens/scheme-motion.css
-   §-1 color-scheme · §0 font · §1 duration · §2 easing (spring linear()) · §2.A ambient-motion · §2.B/§2.C · §3 z-index.
+   §-1 color-scheme · §0 font · §1 duration · §2.W/§2.X motion-weight + cartoon-punch ·
+   §2.A ambient-motion · §2.B/§2.C · §3 z-index.
+   The §2 EASING block (spring linear() curves + per-spring duration clocks +
+   goo-flow curves + cubic-bezier cores/aliases) is carved into the adjacent
+   scheme-spring.css to hold the no-god-module 500-line bound; it @imports
+   IMMEDIATELY AFTER this file (same :root cascade slot).
    Carved from the tokens.css cascade root; @import-ed by tokens.css in
    cascade order. Same :root cascade context as its siblings. */
 :root {
@@ -74,6 +79,20 @@
        §1  DURATION
        ═══════════════════════════════════════════════ */
     --duration-instant: 0.1s;
+    /* BC.W-CONTROL-SMOOTH — the named control-press SURFACE clock (the W-GLASS-CAL
+       per-register-clock discipline). An interactive control the user actively pokes
+       (hover-in, press, toggle) wants the iOS quick ~100-120ms surface beat, NOT the
+       0.2s calm card-crossfade beat — at 0.2s the colour/border/shadow easing drags
+       VISIBLY behind the pointer (the §F "super laggy" read). 0.12s brackets the iOS
+       100-120ms quick-control register (apple-ios27.md). `transition-control` (btn.css)
+       reads THIS so the whole control family re-times from one `:root` override, and
+       the surface beat now tracks the press-squish window (the press is `tap-squish`/
+       `useSpringPress`'s coupled transform leg on the spring's own clock — P3 fade-
+       coupled-to-transform). The bezier `--ease-standard` STAYS for the surface legs
+       (a colour cross-fade on a spring wobbles, §6 binding); only the DURATION shortens.
+       The CALM card register keeps `--duration-fast`/`--duration-normal` — a `<Card>` is
+       not an interactive control. */
+    --duration-control: 0.12s;
     --duration-fast: 0.2s;
     --duration-normal: 0.3s;
     --duration-slow: 0.45s;
@@ -131,138 +150,55 @@
     --motion-stagger-relaxed: 120ms;
 
     /* ═══════════════════════════════════════════════
-       §2  EASING — Spring curves via linear()
-
-       Generated from `scripts/regen-spring-tokens.mjs` — DO NOT hand-edit.
-       To tune a spring: edit the PRESETS table in the script and re-run
-       `node scripts/regen-spring-tokens.mjs`. The script invokes
-       `springLinearStops()` from `@mkbabb/keyframes.js` which solves the
-       analytic second-order damped harmonic oscillator and emits CSS
-       `linear()` stops; this is the single source of truth tying each
-       (response, ζ) pair to its serialized curve.
-
-       iOS-canonical (response, dampingFraction) per AL-X4 §3:
-         smooth: (0.50s, ζ=0.86)  — gentle settle, sub-perceptual overshoot ~0.5%
-         snappy: (0.35s, ζ=0.65)  — quick crisp, overshoot ~+6.8%
-         bouncy: (0.50s, ζ=0.45)  — emphatic overshoot ~+20.5%
-         gentle: (0.70s, ζ=1.00)  — patient critically-damped
-         dock:   (0.32s, ζ=0.70)  — dock expand/collapse morph, overshoot ~+4.6%
-
-       Note on `smooth`: at ζ=0.86 the analytic solver produces a peak
-       of 1.00498 (+0.5%) at the 24% slot — mathematically present but
-       sub-perceptual. I keep ζ=0.86 rather than promoting to ζ=1.0
-       because true critical damping is `gentle`'s register, and
-       `smooth` wants a slightly faster visual settle than `gentle`'s
-       patient curve. The 0.5% peak does not read as overshoot to the
-       eye but does narrowly distinguish `smooth` from `gentle`.
-
-       ── EASING DOCTRINE (AX.W52 §6 — which easing fits which job) ──
-       ONE recorded rule so the animation easing stops forking live. This
-       FORMALIZES the already-correct `transitions.css` idiom (enter=spring,
-       exit=bezier) as the house rule every recipe reaches for.
-
-       BB.W-MOTION-CANON — the table embodies ONE principle (Material 3's
-       spatial-vs-effects split, codified in `docs/precepts/motion-canon.md`
-       as P1): a SPATIAL channel (the box MOVES/RESHAPES — position, SIZE,
-       rotation, corners) rides a `--spring-<name>` (overshoot = physicality);
-       an EFFECTS channel (the surface RE-TINTS — color, opacity, box-shadow)
-       rides a bezier `--ease-*` (a colour cross-fade reads as a wobble on a
-       spring). Each row below is tagged [SPATIAL] (spring) or [EFFECTS]
-       (bezier) so the rule is legible, not implicit. The canon doc is the
-       canonical home; this table is the SOURCE it re-states:
-
-         | Transition kind                                  | Tier      | Easing                   |
-         |--------------------------------------------------|-----------|--------------------------|
-         | Surface props (bg, border, color, box-shadow,    | [EFFECTS] | `--ease-standard` (bezier|
-         |   opacity)                                       |           |   — a colour cross-fade  |
-         |                                                  |           |   reads as a wobble on a |
-         |                                                  |           |   spring)                |
-         | Transform — hover/press/active (scale, translate,| [SPATIAL] | `--spring-smooth` (the   |
-         |   rotate)                                         |           |   ONE button/interactive |
-         |                                                  |           |   scale register) or     |
-         |                                                  |           |   `--spring-snappy`      |
-         | Size / morph (dock V↔H, indicator size-          | [SPATIAL] | `--spring-dock` /        |
-         |   correction, reveal source-rect bloom)          |           |   `--spring-snappy`,     |
-         |                                                  |           |   expressed as a         |
-         |                                                  |           |   COMPOSITOR transform   |
-         |                                                  |           |   (P5) — NEVER an        |
-         |                                                  |           |   animated width/height  |
-         | Enter (mount, popover open, dialog in, reveal)   | [SPATIAL] | `--spring-bouncy` /      |
-         |                                                  |           |   `--spring-snappy`      |
-         | Bold decelerating arrival (the SOTA arrival —    | [SPATIAL] | `--ease-out-expo`        |
-         |   reveal draw-on, big-distance settle)           |           |   (cubic-bezier(.16,1,   |
-         |                                                  |           |   .3,1); the house expo) |
-         | Exit (unmount, close)                            | [EFFECTS] | `--ease-out` /           |
-         |                                                  |           |   `--ease-standard`      |
-         |                                                  |           |   (bezier, NO overshoot —|
-         |                                                  |           |   an exit must never     |
-         |                                                  |           |   overshoot past gone)   |
-         | Position-tracked (specular pointer follow)       | [EFFECTS] | `--ease-standard`        |
-
-       The SOTA arrival ease is the EXISTING `--ease-out-expo` (alias →
-       `--motion-ease-out-expo` = `cubic-bezier(0.16, 1, 0.3, 1)`, the house
-       expo with the value.js `easeOutExpo` JS twin in the MOTION_CURVES table)
-       — NOT a net-new token. The motion canon NAMES it as the arrival register;
-       it is not re-minted (a duplicate alias would fail proof:animation-
-       coherence's EASING-TABLE-BOUND arm — the curve already has its canonical
-       + alias row in `curves.ts`).
-
-       The BUTTON family (`.btn-pill` / `.tap-squish` / `.glass-btn` /
-       `btn-interactive`) UNIFIES its `scale` transform leg onto ONE register —
-       `--spring-smooth` — so hover + press read as one motion language (the
-       hover lift is the restrained `--scale-hover-btn` = 1.05, not the 1.08
-       `--scale-hover`). Springs read alive on TRANSFORM and ENTER (overshoot =
-       physicality); a spring on a COLOR or an EXIT reads as a wobble/hesitation.
-       `--ease-apple` is the ambient-only register (consumed solely by the Pulse
-       aura `--animate-ambient-pulse-easing` — NOT a second smooth-ease authority
-       for interactive transforms; those use `--spring-smooth`).
-       ═══════════════════════════════════════════════ */
-    --spring-smooth: linear(0, 0.09804 2.041%, 0.29288 4.082%, 0.49447 6.122%, 0.66436 8.163%, 0.79208 10.204%, 0.88050 12.245%, 0.93754 14.286%, 0.97182 16.327%, 0.99083 18.367%, 1.00027 20.408%, 1.00413 22.449%, 1.00502 24.490%, 1.00453 26.531%, 1.00355 28.571%, 1.00254 30.612%, 1.00169 32.653%, 1.00105 34.694%, 1.00060 36.735%, 1.00031 38.776%, 1.00014 40.816%, 1.00005 42.857%, 1.00000 44.898%, 1.00000 46.939%, 1.00000 48.980%, 1.00000 51.020%, 1.00000 53.061%, 1.00000 55.102%, 1.00000 57.143%, 1.00000 59.184%, 1.00000 61.224%, 1.00000 63.265%, 1.00000 65.306%, 1.00000 67.347%, 1.00000 69.388%, 1.00000 71.429%, 1.00000 73.469%, 1.00000 75.510%, 1.00000 77.551%, 1.00000 79.592%, 1.00000 81.633%, 1.00000 83.673%, 1.00000 85.714%, 1.00000 87.755%, 1.00000 89.796%, 1.00000 91.837%, 1.00000 93.878%, 1.00000 95.918%, 1.00000 97.959%, 1);
-    --spring-snappy: linear(0, 0.10438 2.041%, 0.32622 4.082%, 0.56668 6.122%, 0.77153 8.163%, 0.91961 10.204%, 1.01073 12.245%, 1.05549 14.286%, 1.06804 16.327%, 1.06171 18.367%, 1.04686 20.408%, 1.03044 22.449%, 1.01629 24.490%, 1.00596 26.531%, 0.99954 28.571%, 0.99633 30.612%, 0.99537 32.653%, 0.99575 34.694%, 0.99674 36.735%, 0.99786 38.776%, 0.99884 40.816%, 0.99956 42.857%, 1.00001 44.898%, 1.00024 46.939%, 1.00000 48.980%, 1.00000 51.020%, 1.00000 53.061%, 1.00000 55.102%, 1.00000 57.143%, 1.00000 59.184%, 1.00000 61.224%, 1.00000 63.265%, 1.00000 65.306%, 1.00000 67.347%, 1.00000 69.388%, 1.00000 71.429%, 1.00000 73.469%, 1.00000 75.510%, 1.00000 77.551%, 1.00000 79.592%, 1.00000 81.633%, 1.00000 83.673%, 1.00000 85.714%, 1.00000 87.755%, 1.00000 89.796%, 1.00000 91.837%, 1.00000 93.878%, 1.00000 95.918%, 1.00000 97.959%, 1);
-    --spring-bouncy: linear(0, 0.11105 2.041%, 0.36430 4.082%, 0.65496 6.122%, 0.90929 8.163%, 1.08828 10.204%, 1.18289 12.245%, 1.20482 14.286%, 1.17641 16.327%, 1.12214 18.367%, 1.06276 20.408%, 1.01239 22.449%, 0.97809 24.490%, 0.96099 26.531%, 0.95826 28.571%, 0.96514 30.612%, 0.97666 32.653%, 0.98873 34.694%, 0.99867 36.735%, 1.00520 38.776%, 1.00825 40.816%, 1.00847 42.857%, 1.00686 44.898%, 1.00443 46.939%, 1.00199 48.980%, 1.00004 51.020%, 0.99880 53.061%, 0.99827 55.102%, 0.99829 57.143%, 0.99866 59.184%, 0.99916 61.224%, 0.99966 63.265%, 1.00004 65.306%, 1.00027 67.347%, 1.00000 69.388%, 1.00000 71.429%, 1.00000 73.469%, 1.00000 75.510%, 1.00000 77.551%, 1.00000 79.592%, 1.00000 81.633%, 1.00000 83.673%, 1.00000 85.714%, 1.00000 87.755%, 1.00000 89.796%, 1.00000 91.837%, 1.00000 93.878%, 1.00000 95.918%, 1.00000 97.959%, 1);
-    --spring-gentle: linear(0, 0.09415 2.041%, 0.27374 4.082%, 0.45506 6.122%, 0.60779 8.163%, 0.72570 10.204%, 0.81213 12.245%, 0.87336 14.286%, 0.91570 16.327%, 0.94445 18.367%, 0.96371 20.408%, 0.97645 22.449%, 0.98481 24.490%, 0.99025 26.531%, 0.99377 28.571%, 0.99604 30.612%, 0.99749 32.653%, 0.99841 34.694%, 0.99900 36.735%, 0.99937 38.776%, 0.99961 40.816%, 0.99975 42.857%, 0.99985 44.898%, 1.00000 46.939%, 1.00000 48.980%, 1.00000 51.020%, 1.00000 53.061%, 1.00000 55.102%, 1.00000 57.143%, 1.00000 59.184%, 1.00000 61.224%, 1.00000 63.265%, 1.00000 65.306%, 1.00000 67.347%, 1.00000 69.388%, 1.00000 71.429%, 1.00000 73.469%, 1.00000 75.510%, 1.00000 77.551%, 1.00000 79.592%, 1.00000 81.633%, 1.00000 83.673%, 1.00000 85.714%, 1.00000 87.755%, 1.00000 89.796%, 1.00000 91.837%, 1.00000 93.878%, 1.00000 95.918%, 1.00000 97.959%, 1);
-    --spring-dock: linear(0, 0.10282 2.041%, 0.31773 4.082%, 0.54783 6.122%, 0.74309 8.163%, 0.88550 10.204%, 0.97598 12.245%, 1.02452 14.286%, 1.04368 16.327%, 1.04501 18.367%, 1.03739 20.408%, 1.02681 22.449%, 1.01672 24.490%, 1.00873 26.531%, 1.00323 28.571%, 0.99996 30.612%, 0.99837 32.653%, 0.99789 34.694%, 0.99805 36.735%, 0.99848 38.776%, 0.99897 40.816%, 0.99940 42.857%, 0.99972 44.898%, 0.99993 46.939%, 1.00004 48.980%, 1.00000 51.020%, 1.00000 53.061%, 1.00000 55.102%, 1.00000 57.143%, 1.00000 59.184%, 1.00000 61.224%, 1.00000 63.265%, 1.00000 65.306%, 1.00000 67.347%, 1.00000 69.388%, 1.00000 71.429%, 1.00000 73.469%, 1.00000 75.510%, 1.00000 77.551%, 1.00000 79.592%, 1.00000 81.633%, 1.00000 83.673%, 1.00000 85.714%, 1.00000 87.755%, 1.00000 89.796%, 1.00000 91.837%, 1.00000 93.878%, 1.00000 95.918%, 1.00000 97.959%, 1);
-
-    /* ── PER-SPRING DURATION CLOCK (BA.W-GLASS-CAL Unit 3) ──
-       Generated from `scripts/regen-spring-tokens.mjs` — DO NOT hand-edit.
-       Each `--spring-<name>-duration` is the spring's OWN 2%-band envelope-settle
-       time `t_s = -ln(0.02) / (ζ·ωₙ)` (ωₙ = 2π/response), derived from the SAME
-       (response, ζ) PRESETS table the `--spring-*` `linear()` curves above
-       derive from — never a hand value.
-
-       WHY: the `--spring-<name>` curve is NORMALIZED to 0..1 and discards the
-       spring's settle time. Pairing it with a generic `--duration-*` clock
-       (0.2/0.3/0.45s) re-timed EVERY spring to the same wall clock — snappy
-       (response 0.35s) and smooth (0.5s) both ran 300ms, so the crisp snappy
-       register dragged a dead sub-pixel tail (the R10-2 read: within-1px at
-       ~109ms but the transition ran 300ms). These clocks re-time UNDER each
-       register to its own settle WITHOUT touching the §6 register canon (which
-       spring fits which job): a `transition` pairs `--spring-<name>` with the
-       matching `--spring-<name>-duration`. */
-    --spring-smooth-duration: 0.36s;
-    --spring-snappy-duration: 0.34s;
-    --spring-bouncy-duration: 0.69s;
-    --spring-gentle-duration: 0.44s;
-    --spring-dock-duration: 0.28s;
-
-    /* Core cubic-bezier (exits, fallbacks, non-spring uses) */
-    --motion-ease-standard: cubic-bezier(0.4, 0, 0.2, 1);
-    --motion-ease-out: cubic-bezier(0, 0, 0.2, 1);
-    --motion-ease-in: cubic-bezier(0.4, 0, 1, 1);
-    --motion-ease-out-expo: cubic-bezier(0.16, 1, 0.3, 1);
-
-    --ease-standard: var(--motion-ease-standard);
-    --ease-out: var(--motion-ease-out);
-    --ease-in: var(--motion-ease-in);
-    --ease-out-expo: var(--motion-ease-out-expo);
-
-    /* Apple-inspired — micro-interactions, switches, dock hover */
-    --motion-ease-apple: cubic-bezier(0.25, 0.1, 0.25, 1);
-
-    --ease-apple: var(--motion-ease-apple);
-    --ease-spring: var(--spring-snappy);
-    --ease-decelerate: var(--ease-out);
-    --ease-accelerate: var(--ease-in);
+       §2.W BD.W-MOTION-WEIGHT — the GOVERNING motion scalar (rest decl)
+
+       The universal driver-motion governor (the @property reg lives in
+       property-regs.css §18 beside --ui-scale). 0 = still, 1 = full cel. Rest
+       1/φ ≈ 0.618: present, alive, never manic (§L4/§L6). DRIVER-scoped: a surface
+       the finger/route moved INHERITS this; an OBSERVER surface (auto-carousel
+       content, list reorder under scroll) pins it to 0 LOCALLY. It co-scales squash
+       depth, overshoot share, anticipation pull-back, and cartoon-shadow travel
+       together so they read as ONE proportioned deformation.
+
+       The SQUISH CAP is derived SITE-LOCALLY at the consuming element (the JS
+       getters in useDockOrientationMorph / useTabIndicator / useTabDragMorph /
+       useLiquidPress read `--motion-weight` off their OWN element and return
+       `1 + (cap_token − 1)·(weight + (1−weight)·flexVel)/0.618`), NEVER a `:root`
+       calc token — an unregistered `:root` cap calc over var(--motion-weight) is
+       substituted at :root (0.618) and inherits the FROZEN text, so a child's
+       scoped weight→0 would never re-evaluate it (the spike's load-bearing
+       correction). The cap reads the EXISTING `--*-max-stretch` token directly with
+       a (weight/0.618) factor — zero new `--*-stretch-k` coefficient cohort. */
+    --motion-weight: 0.618;
+
+    /* ═══════════════════════════════════════════════
+       §2.X BD.W-CARTOON-PUNCH — the third pole, a RAW shaped linear()
+
+       A single damped spring is monotone-from-one-side — it CANNOT dip below origin
+       then overshoot past it. iOS-27's loudest moments want ANTICIPATION (pull back
+       before launch) AND EXAGGERATION (punch past the spring fence). So it ships as
+       a hand-shaped linear() raw token: dip ~−4% BELOW origin (anticipation — a
+       thing no damped spring can do), cross 1.0, punch to ~1.22 (deliberately past
+       the ≤12% spring overshoot fence — which is WHY it is a register, not a
+       spring), settle. Exactly ONE sign change before the peak + a monotone settle
+       after (the shape-sanity fence).
+
+       NOT generated by regen-spring-tokens.mjs (which emits only (response,ζ)
+       springs from springPresets.ts), NOT a SPRING_PRESETS row, NOT a MOTION_CURVES
+       entry (MotionCurveKind is the closed spring|bezier union; a shaped linear()
+       with a negative leg is neither). A raw --ease-* property — zero engine
+       extension, zero TS change. Drives TRANSFORM/opacity ONLY (compositor-safe,
+       both engines — Baseline 17.2+, plain linear(), no -webkit prefix). The surface
+       composing it carries `--motion-weight: 1` in the SAME recipe (the loud-register
+       coupling — so the punch is never accidentally muted to ~+13.6% by inheriting
+       the rest 0.618). PRM zeroes the weight → the amplitude collapses (the springs
+       keep their own respectReducedMotion endpoint-snap). */
+    --ease-cartoon-punch: linear(
+        0, -0.012, -0.038 12%, -0.018 18%,
+        0.18 34%, 0.62 46%, 0.98 56%,
+        1.14 64%, 1.22 70%,
+        1.16 78%, 1.06 86%, 1.012 92%, 1
+    );
 
     /* ═══════════════════════════════════════════════
        §2.A AMBIENT MOTION — Pulse aura tokens
@@ -388,3 +324,40 @@
     --z-debug: 99999;
 
 }
+
+/* ═══════════════════════════════════════════════
+   BD.W-MOTION-WEIGHT — the driver / observer carve (design.md §L4, gate C8)
+
+   The governing weight rests at the 1/φ centroid. LOUD DRIVERS — the dock morph
+   stage, an earned celebration — pin it to 1 so even a SLOW user-driven morph
+   rides the full squish/overshoot register (iOS reserves the weighty bounce for
+   what the user drives). CALM OBSERVERS — a programmatic reorder, an auto-advancing
+   carousel — pin it to 0: auto-motion that springs reads cheap. A single
+   element-scoped assignment, the same seam the PRM carve uses, inheriting down. */
+.liquid-stage {
+    --motion-weight: 1;
+}
+:where([data-reorder], [data-autoplay]) {
+    --motion-weight: 0;
+}
+
+/* ═══════════════════════════════════════════════
+   BD.W-MOTION-WEIGHT / BD.W-CARTOON-PUNCH — the PRM carve
+
+   ONE assignment zeroes ALL of the goo that rides on top of the springs. Because
+   every squish/punch/shadow-travel reads through `--motion-weight` (the site-local
+   cap collapses to `1 + (cap−1)·0 == 1.0`, the inert-cast travel/spread collapse to
+   their rest, the §2c velocity boost is gated by the same weight), this single line
+   zeroes the squash, the overshoot share, the anticipation pull, the cartoon-shadow
+   travel, AND the velocity boost together. The springs keep their own
+   `respectReducedMotion` endpoint-snap in SpringProgress; the weight zeroes only the
+   GOO. The fade survives. `--ease-cartoon-punch` re-aliases to the non-overshooting
+   standard ease (belt-and-suspenders — the weight=0 already kills the punch
+   amplitude on a weight-scaled surface; the alias also removes the curve shape for
+   any consumer driving a non-amplitude leg). */
+@media (prefers-reduced-motion: reduce) {
+    :root {
+        --motion-weight: 0;
+        --ease-cartoon-punch: var(--ease-standard);
+    }
+}

package/dist/styles/tokens/scheme-spring.css

@@ -0,0 +1,239 @@
+/* tokens/scheme-spring.css
+   §2 EASING — Spring curves via linear() · per-spring duration clocks ·
+   goo-morph dwell-flow curves · cubic-bezier cores + aliases.
+
+   Carved from scheme-motion.css's §2 EASING block to hold the no-god-module
+   500-line bound (BD.W-CUT re-grew scheme-motion past it). @import-ed by
+   tokens.css IMMEDIATELY AFTER scheme-motion.css — adjacent same-selector
+   :root{} at the SAME cascade slot, so the spring/easing tokens resolve in the
+   same order their consumers (color-radius and the later partials) expect.
+   Custom properties resolve order-independently within :root, so the split is
+   byte-isomorphic at the cascade. */
+:root {
+    /* ═══════════════════════════════════════════════
+       §2  EASING — Spring curves via linear()
+
+       Generated from `scripts/regen-spring-tokens.mjs` — DO NOT hand-edit.
+       To tune a spring: edit the PRESETS table in the script and re-run
+       `node scripts/regen-spring-tokens.mjs`. The script invokes
+       `springLinearStops()` from `@mkbabb/keyframes.js` which solves the
+       analytic second-order damped harmonic oscillator and emits CSS
+       `linear()` stops; this is the single source of truth tying each
+       (response, ζ) pair to its serialized curve.
+
+       iOS-canonical (response, dampingFraction) per AL-X4 §3 (BC.W-SPRING-EASE
+       eased `snappy`+`bouncy` to fill their clocks + minted the `press` register):
+         smooth: (0.50s, ζ=0.86)  — gentle settle, sub-perceptual overshoot ~0.5%
+         snappy: (0.42s, ζ=0.78)  — quick crisp, eased to fill its clock, overshoot ~+2.0%
+         bouncy: (0.50s, ζ=0.55)  — emphatic, overshoot ~+12.6% (the Apple 12-18% band)
+         gentle: (0.70s, ζ=1.00)  — patient critically-damped
+         dock:   (0.56s, ζ=0.58)  — the WEIGHTY iOS-27 gooey morph (collapse/expand + V↔H + fission), slow inertial settle, audacious overshoot ~+11%
+         press:  (0.15s, ζ=0.86)  — the iOS interactive tap-press, sub-100ms answer, ~+0.5%
+
+       Note on `smooth`: at ζ=0.86 the analytic solver produces a peak
+       of 1.00498 (+0.5%) at the 24% slot — mathematically present but
+       sub-perceptual. I keep ζ=0.86 rather than promoting to ζ=1.0
+       because true critical damping is `gentle`'s register, and
+       `smooth` wants a slightly faster visual settle than `gentle`'s
+       patient curve. The 0.5% peak does not read as overshoot to the
+       eye but does narrowly distinguish `smooth` from `gentle`.
+
+       ── EASING DOCTRINE (AX.W52 §6 — which easing fits which job) ──
+       ONE recorded rule so the animation easing stops forking live. This
+       FORMALIZES the already-correct `transitions.css` idiom (enter=spring,
+       exit=bezier) as the house rule every recipe reaches for.
+
+       BB.W-MOTION-CANON — the table embodies ONE principle (Material 3's
+       spatial-vs-effects split, codified in `docs/precepts/motion-canon.md`
+       as P1): a SPATIAL channel (the box MOVES/RESHAPES — position, SIZE,
+       rotation, corners) rides a `--spring-<name>` (overshoot = physicality);
+       an EFFECTS channel (the surface RE-TINTS — color, opacity, box-shadow)
+       rides a bezier `--ease-*` (a colour cross-fade reads as a wobble on a
+       spring). Each row below is tagged [SPATIAL] (spring) or [EFFECTS]
+       (bezier) so the rule is legible, not implicit. The canon doc is the
+       canonical home; this table is the SOURCE it re-states:
+
+         | Transition kind                                  | Tier      | Easing                   |
+         |--------------------------------------------------|-----------|--------------------------|
+         | Surface props (bg, border, color, box-shadow,    | [EFFECTS] | `--ease-standard` (bezier|
+         |   opacity)                                       |           |   — a colour cross-fade  |
+         |                                                  |           |   reads as a wobble on a |
+         |                                                  |           |   spring)                |
+         | Transform — hover/press/active (scale, translate,| [SPATIAL] | `--spring-smooth` (the   |
+         |   rotate)                                         |           |   ONE button/interactive |
+         |                                                  |           |   scale register) or     |
+         |                                                  |           |   `--spring-snappy`      |
+         | Size / morph (dock V↔H, indicator size-          | [SPATIAL] | `--spring-dock` /        |
+         |   correction, reveal source-rect bloom)          |           |   `--spring-snappy`,     |
+         |                                                  |           |   expressed as a         |
+         |                                                  |           |   COMPOSITOR transform   |
+         |                                                  |           |   (P5) — NEVER an        |
+         |                                                  |           |   animated width/height  |
+         | Enter (mount, popover open, dialog in, reveal)   | [SPATIAL] | `--spring-bouncy` /      |
+         |                                                  |           |   `--spring-snappy`      |
+         | Bold decelerating arrival (the SOTA arrival —    | [SPATIAL] | `--ease-out-expo`        |
+         |   reveal draw-on, big-distance settle)           |           |   (cubic-bezier(.16,1,   |
+         |                                                  |           |   .3,1); the house expo) |
+         | Exit (unmount, close)                            | [EFFECTS] | `--ease-out` /           |
+         |                                                  |           |   `--ease-standard`      |
+         |                                                  |           |   (bezier, NO overshoot —|
+         |                                                  |           |   an exit must never     |
+         |                                                  |           |   overshoot past gone)   |
+         | Position-tracked (specular pointer follow)       | [EFFECTS] | `--ease-standard`        |
+
+       The SOTA arrival ease is the EXISTING `--ease-out-expo` (alias →
+       `--motion-ease-out-expo` = `cubic-bezier(0.16, 1, 0.3, 1)`, the house
+       expo with the value.js `easeOutExpo` JS twin in the MOTION_CURVES table)
+       — NOT a net-new token. The motion canon NAMES it as the arrival register;
+       it is not re-minted (a duplicate alias would fail proof:animation-
+       coherence's EASING-TABLE-BOUND arm — the curve already has its canonical
+       + alias row in `curves.ts`).
+
+       The BUTTON family (`.btn-pill` / `.tap-squish` / `.glass-btn` /
+       `btn-interactive`) UNIFIES its `scale` transform leg onto ONE register —
+       `--spring-smooth` — so hover + press read as one motion language (the
+       hover lift is the restrained `--scale-hover-btn` = 1.05, not the 1.08
+       `--scale-hover`). Springs read alive on TRANSFORM and ENTER (overshoot =
+       physicality); a spring on a COLOR or an EXIT reads as a wobble/hesitation.
+       `--ease-apple` is the ambient-only register (consumed solely by the Pulse
+       aura `--animate-ambient-pulse-easing` — NOT a second smooth-ease authority
+       for interactive transforms; those use `--spring-smooth`).
+       ═══════════════════════════════════════════════ */
+    --spring-smooth: linear(0, 0.09979 2.041%, 0.30180 4.082%, 0.51333 6.122%, 0.69190 8.163%, 0.82461 10.204%, 0.91382 12.245%, 0.96822 14.286%, 0.99774 16.327%, 1.01113 18.367%, 1.01507 20.408%, 1.01416 22.449%, 1.01127 24.490%, 1.00802 26.531%, 1.00518 28.571%, 1.00302 30.612%, 1.00154 32.653%, 1.00062 34.694%, 1.00010 36.735%, 0.99986 38.776%, 1.00000 40.816%, 1.00000 42.857%, 1.00000 44.898%, 1.00000 46.939%, 1.00000 48.980%, 1.00000 51.020%, 1.00000 53.061%, 1.00000 55.102%, 1.00000 57.143%, 1.00000 59.184%, 1.00000 61.224%, 1.00000 63.265%, 1.00000 65.306%, 1.00000 67.347%, 1.00000 69.388%, 1.00000 71.429%, 1.00000 73.469%, 1.00000 75.510%, 1.00000 77.551%, 1.00000 79.592%, 1.00000 81.633%, 1.00000 83.673%, 1.00000 85.714%, 1.00000 87.755%, 1.00000 89.796%, 1.00000 91.837%, 1.00000 93.878%, 1.00000 95.918%, 1.00000 97.959%, 1);
+    --spring-snappy: linear(0, 0.10159 2.041%, 0.31119 4.082%, 0.53355 6.122%, 0.72176 8.163%, 0.86005 10.204%, 0.95002 12.245%, 1.00115 14.286%, 1.02488 16.327%, 1.03150 18.367%, 1.02891 20.408%, 1.02249 22.449%, 1.01542 24.490%, 1.00934 26.531%, 1.00480 28.571%, 1.00181 30.612%, 1.00009 32.653%, 0.99926 34.694%, 0.99901 36.735%, 0.99907 38.776%, 0.99927 40.816%, 0.99949 42.857%, 0.99969 44.898%, 0.99984 46.939%, 0.99993 48.980%, 0.99999 51.020%, 1.00000 53.061%, 1.00000 55.102%, 1.00000 57.143%, 1.00000 59.184%, 1.00000 61.224%, 1.00000 63.265%, 1.00000 65.306%, 1.00000 67.347%, 1.00000 69.388%, 1.00000 71.429%, 1.00000 73.469%, 1.00000 75.510%, 1.00000 77.551%, 1.00000 79.592%, 1.00000 81.633%, 1.00000 83.673%, 1.00000 85.714%, 1.00000 87.755%, 1.00000 89.796%, 1.00000 91.837%, 1.00000 93.878%, 1.00000 95.918%, 1.00000 97.959%, 1);
+    --spring-bouncy: linear(0, 0.10599 2.041%, 0.33511 4.082%, 0.58672 6.122%, 0.80212 8.163%, 0.95655 10.204%, 1.04836 12.245%, 1.08867 14.286%, 1.09341 16.327%, 1.07802 18.367%, 1.05470 20.408%, 1.03159 22.449%, 1.01303 24.490%, 1.00049 26.531%, 0.99363 28.571%, 0.99114 30.612%, 0.99150 32.653%, 0.99333 34.694%, 0.99560 36.735%, 0.99767 38.776%, 0.99924 40.816%, 1.00023 42.857%, 1.00072 44.898%, 1.00000 46.939%, 1.00000 48.980%, 1.00000 51.020%, 1.00000 53.061%, 1.00000 55.102%, 1.00000 57.143%, 1.00000 59.184%, 1.00000 61.224%, 1.00000 63.265%, 1.00000 65.306%, 1.00000 67.347%, 1.00000 69.388%, 1.00000 71.429%, 1.00000 73.469%, 1.00000 75.510%, 1.00000 77.551%, 1.00000 79.592%, 1.00000 81.633%, 1.00000 83.673%, 1.00000 85.714%, 1.00000 87.755%, 1.00000 89.796%, 1.00000 91.837%, 1.00000 93.878%, 1.00000 95.918%, 1.00000 97.959%, 1);
+    --spring-gentle: linear(0, 0.09415 2.041%, 0.27374 4.082%, 0.45506 6.122%, 0.60779 8.163%, 0.72570 10.204%, 0.81213 12.245%, 0.87336 14.286%, 0.91570 16.327%, 0.94445 18.367%, 0.96371 20.408%, 0.97645 22.449%, 0.98481 24.490%, 0.99025 26.531%, 0.99377 28.571%, 0.99604 30.612%, 0.99749 32.653%, 0.99841 34.694%, 0.99900 36.735%, 0.99937 38.776%, 0.99961 40.816%, 0.99975 42.857%, 0.99985 44.898%, 1.00000 46.939%, 1.00000 48.980%, 1.00000 51.020%, 1.00000 53.061%, 1.00000 55.102%, 1.00000 57.143%, 1.00000 59.184%, 1.00000 61.224%, 1.00000 63.265%, 1.00000 65.306%, 1.00000 67.347%, 1.00000 69.388%, 1.00000 71.429%, 1.00000 73.469%, 1.00000 75.510%, 1.00000 77.551%, 1.00000 79.592%, 1.00000 81.633%, 1.00000 83.673%, 1.00000 85.714%, 1.00000 87.755%, 1.00000 89.796%, 1.00000 91.837%, 1.00000 93.878%, 1.00000 95.918%, 1.00000 97.959%, 1);
+    --spring-dock: linear(0, 0.10470 2.041%, 0.32797 4.082%, 0.57059 6.122%, 0.77747 8.163%, 0.92676 10.204%, 1.01801 12.245%, 1.06194 14.286%, 1.07303 16.327%, 1.06501 18.367%, 1.04857 20.408%, 1.03088 22.449%, 1.01588 24.490%, 1.00510 26.531%, 0.99855 28.571%, 0.99542 30.612%, 0.99467 32.653%, 0.99528 34.694%, 0.99649 36.735%, 0.99778 38.776%, 0.99887 40.816%, 0.99965 42.857%, 1.00012 44.898%, 1.00034 46.939%, 1.00000 48.980%, 1.00000 51.020%, 1.00000 53.061%, 1.00000 55.102%, 1.00000 57.143%, 1.00000 59.184%, 1.00000 61.224%, 1.00000 63.265%, 1.00000 65.306%, 1.00000 67.347%, 1.00000 69.388%, 1.00000 71.429%, 1.00000 73.469%, 1.00000 75.510%, 1.00000 77.551%, 1.00000 79.592%, 1.00000 81.633%, 1.00000 83.673%, 1.00000 85.714%, 1.00000 87.755%, 1.00000 89.796%, 1.00000 91.837%, 1.00000 93.878%, 1.00000 95.918%, 1.00000 97.959%, 1);
+    --spring-press: linear(0, 0.09979 2.041%, 0.30180 4.082%, 0.51333 6.122%, 0.69190 8.163%, 0.82461 10.204%, 0.91382 12.245%, 0.96822 14.286%, 0.99774 16.327%, 1.01113 18.367%, 1.01507 20.408%, 1.01416 22.449%, 1.01127 24.490%, 1.00802 26.531%, 1.00518 28.571%, 1.00302 30.612%, 1.00154 32.653%, 1.00062 34.694%, 1.00010 36.735%, 0.99986 38.776%, 0.99978 40.816%, 0.99978 42.857%, 0.99982 44.898%, 0.99987 46.939%, 0.99991 48.980%, 0.99995 51.020%, 0.99997 53.061%, 1.00000 55.102%, 1.00000 57.143%, 1.00000 59.184%, 1.00000 61.224%, 1.00000 63.265%, 1.00000 65.306%, 1.00000 67.347%, 1.00000 69.388%, 1.00000 71.429%, 1.00000 73.469%, 1.00000 75.510%, 1.00000 77.551%, 1.00000 79.592%, 1.00000 81.633%, 1.00000 83.673%, 1.00000 85.714%, 1.00000 87.755%, 1.00000 89.796%, 1.00000 91.837%, 1.00000 93.878%, 1.00000 95.918%, 1.00000 97.959%, 1);
+    --spring-timeline-head: linear(0, 0.10159 2.041%, 0.31119 4.082%, 0.53355 6.122%, 0.72176 8.163%, 0.86005 10.204%, 0.95002 12.245%, 1.00115 14.286%, 1.02488 16.327%, 1.03150 18.367%, 1.02891 20.408%, 1.02249 22.449%, 1.01542 24.490%, 1.00934 26.531%, 1.00480 28.571%, 1.00181 30.612%, 1.00009 32.653%, 0.99926 34.694%, 0.99901 36.735%, 0.99907 38.776%, 0.99927 40.816%, 0.99949 42.857%, 0.99969 44.898%, 0.99984 46.939%, 0.99993 48.980%, 0.99999 51.020%, 1.00000 53.061%, 1.00000 55.102%, 1.00000 57.143%, 1.00000 59.184%, 1.00000 61.224%, 1.00000 63.265%, 1.00000 65.306%, 1.00000 67.347%, 1.00000 69.388%, 1.00000 71.429%, 1.00000 73.469%, 1.00000 75.510%, 1.00000 77.551%, 1.00000 79.592%, 1.00000 81.633%, 1.00000 83.673%, 1.00000 85.714%, 1.00000 87.755%, 1.00000 89.796%, 1.00000 91.837%, 1.00000 93.878%, 1.00000 95.918%, 1.00000 97.959%, 1);
+    --spring-timeline-fill: linear(0, 0.09920 2.041%, 0.29878 4.082%, 0.50690 6.122%, 0.68248 8.163%, 0.81346 10.204%, 0.90241 12.245%, 0.95775 14.286%, 0.98897 16.327%, 1.00435 18.367%, 1.01023 20.408%, 1.01098 22.449%, 1.00939 24.490%, 1.00707 26.531%, 1.00484 28.571%, 1.00304 30.612%, 1.00173 32.653%, 1.00087 34.694%, 1.00034 36.735%, 1.00006 38.776%, 0.99992 40.816%, 1.00000 42.857%, 1.00000 44.898%, 1.00000 46.939%, 1.00000 48.980%, 1.00000 51.020%, 1.00000 53.061%, 1.00000 55.102%, 1.00000 57.143%, 1.00000 59.184%, 1.00000 61.224%, 1.00000 63.265%, 1.00000 65.306%, 1.00000 67.347%, 1.00000 69.388%, 1.00000 71.429%, 1.00000 73.469%, 1.00000 75.510%, 1.00000 77.551%, 1.00000 79.592%, 1.00000 81.633%, 1.00000 83.673%, 1.00000 85.714%, 1.00000 87.755%, 1.00000 89.796%, 1.00000 91.837%, 1.00000 93.878%, 1.00000 95.918%, 1.00000 97.959%, 1);
+    --spring-timeline-press: linear(0, 0.10282 2.041%, 0.31773 4.082%, 0.54783 6.122%, 0.74309 8.163%, 0.88550 10.204%, 0.97598 12.245%, 1.02452 14.286%, 1.04368 16.327%, 1.04501 18.367%, 1.03739 20.408%, 1.02681 22.449%, 1.01672 24.490%, 1.00873 26.531%, 1.00323 28.571%, 0.99996 30.612%, 0.99837 32.653%, 0.99789 34.694%, 0.99805 36.735%, 0.99848 38.776%, 0.99897 40.816%, 0.99940 42.857%, 0.99972 44.898%, 0.99993 46.939%, 1.00004 48.980%, 1.00009 51.020%, 1.00000 53.061%, 1.00000 55.102%, 1.00000 57.143%, 1.00000 59.184%, 1.00000 61.224%, 1.00000 63.265%, 1.00000 65.306%, 1.00000 67.347%, 1.00000 69.388%, 1.00000 71.429%, 1.00000 73.469%, 1.00000 75.510%, 1.00000 77.551%, 1.00000 79.592%, 1.00000 81.633%, 1.00000 83.673%, 1.00000 85.714%, 1.00000 87.755%, 1.00000 89.796%, 1.00000 91.837%, 1.00000 93.878%, 1.00000 95.918%, 1.00000 97.959%, 1);
+
+    /* ── PER-SPRING DURATION CLOCK (BA.W-GLASS-CAL Unit 3) ──
+       Generated from `scripts/regen-spring-tokens.mjs` — DO NOT hand-edit.
+       Each `--spring-<name>-duration` is the spring's OWN 2%-band envelope-settle
+       time `t_s = -ln(0.02) / (ζ·ωₙ)` (ωₙ = 2π/response), derived from the SAME
+       (response, ζ) PRESETS table the `--spring-*` `linear()` curves above
+       derive from — never a hand value.
+
+       WHY: the `--spring-<name>` curve is NORMALIZED to 0..1 and discards the
+       spring's settle time. Pairing it with a generic `--duration-*` clock
+       (0.2/0.3/0.45s) re-timed EVERY spring to the same wall clock — snappy
+       (response 0.35s) and smooth (0.5s) both ran 300ms, so the crisp snappy
+       register dragged a dead sub-pixel tail (the R10-2 read: within-1px at
+       ~109ms but the transition ran 300ms). These clocks re-time UNDER each
+       register to its own settle WITHOUT touching the §6 register canon (which
+       spring fits which job): a `transition` pairs `--spring-<name>` with the
+       matching `--spring-<name>-duration`. */
+    --spring-smooth-duration: 0.45s;
+    --spring-snappy-duration: 0.4s;
+    --spring-bouncy-duration: 0.62s;
+    --spring-gentle-duration: 0.51s;
+    --spring-dock-duration: 0.66s;
+    --spring-press-duration: 0.16s;
+    --spring-timeline-head-duration: 0.29s;
+    --spring-timeline-fill-duration: 0.35s;
+    --spring-timeline-press-duration: 0.2s;
+
+    /* ── DECK SLIDE-TRANSITION CURVE (BC.W-DECK) ──
+       The full-viewport keyboard-paged deck's slide transition rides the SAME
+       `--spring-smooth` register the deck's `DECK_SPRING` samples (the calm-settle
+       `.smooth` family; the (response, ζ) pair lives in deck/constants.ts) — an
+       ALIAS, NOT a new spring family (the donor's own fence + the W-GLASS-CAL spring
+       fence). `--spring-deck: var(--spring-smooth)` so a deck slide reads as a calm
+       settle, not a bounce; the JS half (`installDeckSpring`) consumes keyframes.js
+       `springTimingFunction(DECK_SPRING)` for the count-up easing. */
+    --spring-deck: var(--spring-smooth);
+
+    /* ── GOO-MORPH SLIDE-TRANSITION FLOW CURVES (BD.W-GOO-CAROUSEL-DECK) ──
+       The per-consumer dwell-flow curves the ONE `useGooMorph` engine reads off its
+       `tokenPrefix` (`--{prefix}-flow`/`-duration`/`-max-stretch`). These are NOT new
+       spring presets — they are the SAME geometry-law DWELL shape the dot-pip worm
+       (`--pager-worm-flow`, defined just below) ships: a curve that rises to ~mid by ~46% of
+       the clock, DWELLS at the midpoint (the FAT metaball NECK held open across the gap),
+       then contracts. The dwell is what makes the morph read as weighty liquid rather than
+       a fast flicker (the W-GLASS-CAL spring fence: a normalized spring `linear()`
+       front-loads — reaches ~0.5 by ~6% REGARDLESS of ζ — so the neck would collapse in
+       ~150ms then sit dead; the dwell curve spreads `--goo-t`'s 0→1 across the REAL clock).
+
+       --carousel-goo-flow: the drag register — the dwell shape with a gentle terminal
+         overshoot (+1.5%, the snappy LAND); the plate necks across the gap then snaps. */
+    --carousel-goo-flow: linear(
+        0, 0.02206 4.167%, 0.07922 8.333%, 0.15799 12.5%, 0.24486 16.667%,
+        0.32632 20.833%, 0.38889 25%, 0.41905 29.167%, 0.43667 33.333%,
+        0.4575 37.5%, 0.47833 41.667%, 0.49917 45.833%, 0.52 50%, 0.54083 54.167%,
+        0.56167 58.333%, 0.58373 62.5%, 0.62943 66.667%, 0.69424 70.833%,
+        0.76859 75%, 0.84373 79.167%, 0.91191 83.333%, 0.96658 87.5%,
+        1.0023 91.667%, 1.01466 95.833%, 1
+    );
+    --carousel-goo-duration: 0.95s; /* the card-plate clock — faster than the dot's 1.8s
+       (a card-plate morph is bigger, reads heavy sooner) yet still slow enough to dwell. */
+    --carousel-goo-max-stretch: 1.24; /* the plate swell — raised (1.16→1.24) for a more
+       decisive welling neck across the gap (JUDGE-2 §3: the merge read SUBTLE). Still
+       LOWER than the dot's 1.45 (a wide plate stretching 1.45 reads as taffy). */
+
+    /* --deck-goo-flow: the calmer presentation register — the SAME dwell shape with NO
+       terminal overshoot (a full-viewport slide must not bounce — the vestibular floor;
+       the curve clamps at 1.0, no >1 stops). */
+    --deck-goo-flow: linear(
+        0, 0.02206 4.167%, 0.07922 8.333%, 0.15799 12.5%, 0.24486 16.667%,
+        0.32632 20.833%, 0.38889 25%, 0.41905 29.167%, 0.43667 33.333%,
+        0.4575 37.5%, 0.47833 41.667%, 0.49917 45.833%, 0.52 50%, 0.54083 54.167%,
+        0.56167 58.333%, 0.58373 62.5%, 0.64 66.667%, 0.71 70.833%,
+        0.78 75%, 0.85 79.167%, 0.91 83.333%, 0.96 87.5%,
+        0.985 91.667%, 0.997 95.833%, 1
+    );
+    --deck-goo-duration: 1.1s; /* the full-viewport register — slow + dignified. */
+    --deck-goo-max-stretch: 1.1; /* the gentlest swell — a big slide necks barely. */
+
+    /* --pager-worm-flow: the dot-pip worm register (BD.W-PAGER-GOO-MORPH) — the SAME
+       geometry-law DWELL shape as the carousel/deck siblings (the metaball NECK held open
+       across the gap), with a gentle terminal overshoot (+1.5%, the bouncy LAND). Homed
+       HERE beside its goo-flow family (the curve-definition home — a `linear()` SHIPS in
+       the §2 EASING register, never serialized inline in the SFC) so the three goo-morph
+       dwell curves live as ONE charted family. NOT a normalized spring: `springLinearStops`
+       front-loads (reaches ~0.5 by ~6% REGARDLESS of ζ), so a `--spring-bouncy` would
+       collapse the neck in ~150ms; this dwell spreads `--worm-t`'s 0→1 across the REAL
+       1.8s clock (the FAT NECK alive ~830ms — JUDGE-1 ≥700ms bar). */
+    --pager-worm-flow: linear(
+        0, 0.02206 4.167%, 0.07922 8.333%, 0.15799 12.5%, 0.24486 16.667%,
+        0.32632 20.833%, 0.38889 25%, 0.41905 29.167%, 0.43667 33.333%,
+        0.4575 37.5%, 0.47833 41.667%, 0.49917 45.833%, 0.52 50%, 0.54083 54.167%,
+        0.56167 58.333%, 0.58373 62.5%, 0.62943 66.667%, 0.69424 70.833%,
+        0.76859 75%, 0.84373 79.167%, 0.91191 83.333%, 0.96658 87.5%,
+        1.0023 91.667%, 1.01466 95.833%, 1
+    );
+    --pager-worm-duration: 1.8s; /* the dot's OWN clock — slow + weighty (motion-canon P4);
+       the flow above dwells across the whole clock so the neck is alive for ~1s. */
+    --pager-worm-max-stretch: 1.45; /* the velocity-swell — a VISIBLE liquid squish (the
+       worm narrows ≈1/1.45≈0.69× cross-axis at peak swell, on top of the floored pinch). */
+
+    /* Core cubic-bezier (exits, fallbacks, non-spring uses) */
+    --motion-ease-standard: cubic-bezier(0.4, 0, 0.2, 1);
+    --motion-ease-out: cubic-bezier(0, 0, 0.2, 1);
+    --motion-ease-in: cubic-bezier(0.4, 0, 1, 1);
+    --motion-ease-out-expo: cubic-bezier(0.16, 1, 0.3, 1);
+
+    --ease-standard: var(--motion-ease-standard);
+    --ease-out: var(--motion-ease-out);
+    --ease-in: var(--motion-ease-in);
+    --ease-out-expo: var(--motion-ease-out-expo);
+
+    /* Apple-inspired — micro-interactions, switches, dock hover */
+    --motion-ease-apple: cubic-bezier(0.25, 0.1, 0.25, 1);
+
+    --ease-apple: var(--motion-ease-apple);
+    --ease-spring: var(--spring-snappy);
+    --ease-decelerate: var(--ease-out);
+    --ease-accelerate: var(--ease-in);
+
+    /* BC.W-MOTION-PRESETS — the brand "convergence-reveal" / partial-sum settle.
+       The Fourier partial sum converges ONTO its target with a decaying Gibbs
+       ripple — a monotone critically-damped approach, NO sustained overshoot past
+       terminal. That curve IS `gentle` (ζ=1.0, overshoot 0), so this is the ALIAS
+       reuse: `--ease-convergence` resolves to `--spring-gentle` (no new
+       SPRING_PRESETS row, no second curve). The settle clock is the per-spring
+       `--spring-gentle-duration`. The N-term build-up is the `useStagger`
+       orchestration ON this curve (a documented recipe, not a new primitive). */
+    --ease-convergence: var(--spring-gentle);
+}

package/dist/styles/tokens/scroll-tokens.css

@@ -41,13 +41,13 @@
     --scroll-cascade-range-end: 45%;
     --scroll-cascade-column-stagger: 60ms;
     /* .scroll-pin — the fixed-stage-advances-time scroll-pinned register. The stage
-       height is the Nvh temporal space the sticky stage advances its phases across;
-       the phase windows are the scroll-progress fractions the stage's internal
-       phases key off (grid-reveal 0..45%, expand 45..90%, settle 90..100% — the
-       Codrops phased-sequence shape). */
+       height is the Nvh temporal space the sticky stage advances its phases across.
+       The phase windows the stage's internal phases key off (grid-reveal 0..reveal-end,
+       expand reveal-end..expand-end, settle expand-end..100% — the Codrops phased-
+       sequence shape) live in scroll-choreography.css as the `-frac` numerators
+       (`--scroll-pin-phase-reveal-end-frac` 0.382 / `--scroll-pin-phase-expand-end-frac`
+       0.9), the form the calc()s consume. */
     --scroll-pin-stage-height: 320vh;
-    --scroll-pin-phase-reveal-end: 45%;
-    --scroll-pin-phase-expand-end: 90%;
     --scroll-pin-lift: 2.5rem;
 
     /* TOP-LAYER (animations.css §TOP-LAYER). The `@starting-style` entry/exit
@@ -82,19 +82,19 @@
        both engine paths are timing-IDENTICAL: the FLIP fallback (via
        `--dock-motion-resize`), the native View-Transition group (its
        `animation-timing-function`), AND the runtime SpringProgress driver
-       (AU.W8.3) consume this same `(0.32, 0.7)` source curve. Retires the AQ.W6
+       (AU.W8.3) consume this same source curve. Retires the AQ.W6
        fork where the VT path ran `--vt-ease` (apple-spring, +27.5% overshoot)
        while the FLIP ran the snappy spring. `--vt-ease` still defaults the
        other VT recipes (`.gl-list-item`).
 
-       AU.W8 — RETARGET snappy → dock. `--spring-snappy` (ζ=0.65, +6.8%) read as
-       mechanical: a stiff plateau at ~48% with too little overshoot to feel
-       alive ("not iOS-smooth"). AW.W2 then retuned the dock spring to the
-       iOS-control register: `--spring-dock` ((0.32, 0.7), ~+4.6%) settles snappy
-       with a restrained overshoot — the felt iOS dock morph, off the prior
-       playful +18.5% register. The runtime driver's DOCK_SPRING const MIRRORS the
-       (0.32, 0.7) PRESETS row, so the CSS token and the JS-driven morph are
-       bit-identical. */
+       BD.W-DOCK-CORE — the WEIGHTY iOS-27 re-tune. `--spring-dock` is now the
+       slow inertial gooey-morph register: (0.56s, ζ=0.58), an audacious overshoot
+       ~+11% then a deliberate settle over the 0.6s analytic clock — the
+       liquid-weight law (collapse/expand + V↔H + fission all share it). The prior
+       tight (0.32, 0.7) near-critical snap (~+4.6%) is retired. The runtime
+       driver's DOCK_SPRING const DERIVES from the `dock` PRESETS row via
+       `springPreset("dock")` (the no-second-authority fence — ONE table row, the
+       CSS token + the JS-driven morph + the const all flow from it). */
     --dock-resize-spring: var(--spring-dock);
 
     /* AT.W7-dock-b — dock-control press micro-feedback spring. A TRANSFORM-ONLY