Struct MotionTokens 

pub struct MotionTokens { /* private fields */ }

The application-global motion policy: every duration, stagger beat, and spring response the crate animates with, resolved through one value.

crate::init installs MotionTokens::DEFAULT unless the application already chose. To customize, build from default() and set the result — before windows render, so no frame is timed under two policies:

gpui_ai::init(cx);
gpui_ai::motion::MotionTokens::default()
    .with_quick(Duration::from_millis(120))
    .set(cx);

Components use the quick role for acknowledgments, standard for entrances and compact disclosures, and spring roles for bounded retargets. Named repeating effects keep their individual tempos. The deliberate role is available for application composition without adding motion implicitly.

What the policy cannot do: override reduced motion. cx.reduce_motion() stays authoritative in every consumer, so customization tunes tempo and feel, never whether travel happens for someone who asked for stillness.

Implementations

impl MotionTokens

pub const DEFAULT: Self

The crate’s default duration, spring, and repeating-effect policy.

pub fn read(cx: &App) -> &Self

The policy the application installed, or the crate’s default where it installed none. Never panics: a window rendered before init runs is timed by Self::DEFAULT, the same values init would install.

pub fn set(self, cx: &mut App)

Makes this policy the application’s, for every component at once.

pub const fn instant(&self) -> Duration

Focus, press acknowledgement, pointer capture: zero, because a direct response to input is not eased.

pub const fn quick(&self) -> Duration

Icon or label swap, copy acknowledgement, close.

pub const fn hover_glide(&self) -> Duration

One hover highlight gliding between uniform rows, chasing the pointer on the strong ease-out. Faster than quick because the highlight follows input rather than acknowledging it; any reduced preference snaps the highlight instead.

pub const fn standard(&self) -> Duration

Selected indicator, compact disclosure, progress retarget — and the crate’s entrance tempo.

pub const fn deliberate(&self) -> Duration

Larger bounded panel or success choreography.

Restricted by contract to modal- and sheet-class surfaces and explanatory composition: nothing smaller — tooltips, popovers, dropdowns, toggles, row state — may bind to it, because 380ms on an element the reader is waiting on reads as lag, not weight. Popover-class entrances bind to quick.

pub const fn press_spring(&self) -> SpringRole

Reversible control compression: fast and fully damped.

pub const fn selection_spring(&self) -> SpringRole

An active pill or marker gliding between targets.

pub const fn disclosure_spring(&self) -> SpringRole

Bounded panel geometry; no bounce, so it also suits destructive and error surfaces.

pub const fn reflow_spring(&self) -> SpringRole

Visible row displacement: critically damped, velocity-preserving.

pub const fn preference(&self) -> MotionPreference

The policy’s own motion preference, before the OS signal composes.

pub fn effective_preference(cx: &App) -> MotionPreference

The preference actually in effect: the policy’s, floored at MotionPreference::Crossfade while the OS asks for reduced motion. Every clock, spring, and loop in the crate resolves through this.

pub const fn with_preference(self, preference: MotionPreference) -> Self

Replaces the preference.

pub const fn with_instant(self, duration: Duration) -> Self

Replaces the instant duration.

pub const fn with_quick(self, duration: Duration) -> Self

Replaces the quick duration.

pub const fn with_hover_glide(self, duration: Duration) -> Self

Replaces the hover_glide duration.

pub const fn with_standard(self, duration: Duration) -> Self

Replaces the standard duration, which is also the entrance tempo every reveal in the crate settles at.

pub const fn with_deliberate(self, duration: Duration) -> Self

Replaces the deliberate duration.

pub const fn with_stagger_beat(self, beat: Duration) -> Self

Replaces the per-item stagger beat. The participation and aggregate caps are policy, not configuration, and stay where they are.

pub const fn with_press_spring(self, response: Duration, damping: f32) -> Self

Replaces the press spring’s feel.

pub const fn with_selection_spring( self, response: Duration, damping: f32, ) -> Self

Replaces the selection spring’s feel.

pub const fn with_disclosure_spring( self, response: Duration, damping: f32, ) -> Self

Replaces the disclosure spring’s feel.

pub const fn with_reflow_spring(self, response: Duration, damping: f32) -> Self

Replaces the reflow spring’s feel.

pub fn arrival_stagger(&self, index: usize, arriving: usize) -> Duration

The delay item index of arriving newly visible items waits before its arrival plays — the stagger role.

The cascade decelerates: delays are spread with ease-out spacing, so later items land closer together and the group resolves quickly instead of trailing. At most six items take part, the whole cascade spends at most 220 ms, and items past the participation bound arrive with the last beat rather than after it — the bounded-choreography caps, which are policy rather than configuration. The per-item beat scales with with_stagger_beat.

Provisional like the other unconsumed roles: the motion lab is its first consumer and its validator, and the shipped reveal_staggered keeps its documented linear beats regardless.

Trait Implementations

impl Clone for MotionTokens

fn clone(&self) -> MotionTokens

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for MotionTokens

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for MotionTokens

fn default() -> Self

Returns the “default value” for a type.

impl Global for MotionTokens

impl PartialEq for MotionTokens

fn eq(&self, other: &MotionTokens) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for MotionTokens