Struct PositionJitter 

pub struct PositionJitter {
    pub width: Option<f64>,
    pub height: Option<f64>,
    pub seed: u64,
}

Add random noise to x and y positions to reduce overplotting.

Each value moves by a uniform amount in [-width, width) (x) and [-height, height) (y). None (the default) means “40% of the data’s resolution”, as in ggplot2: the smallest gap between distinct values, or 1 for integer/constant data — so integer-coded categories (1, 2, 3, …) get ±0.4 and finely-spaced continuous data gets proportionally small noise. Some(0.0) disables jitter along that axis.

Fields

width: Option<f64>

Horizontal jitter amount in data units; None = 0.4 × resolution(x).

height: Option<f64>

Vertical jitter amount in data units; None = 0.4 × resolution(y).

seed: u64

Seed of the deterministic noise stream.

Implementations

impl PositionJitter

pub fn new(width: f64, height: f64) -> Self

Explicit jitter amounts (data units), default seed.

pub fn with_width(self, width: f64) -> Self

Set the horizontal jitter amount (data units).

pub fn with_height(self, height: f64) -> Self

Set the vertical jitter amount (data units).

pub fn with_seed(self, seed: u64) -> Self

Use a different seed (ggplot2’s position_jitter(seed = …)).

Trait Implementations

impl Clone for PositionJitter

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for PositionJitter

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

Formats the value using the given formatter.

impl Default for PositionJitter

fn default() -> Self

Returns the “default value” for a type.

impl Position for PositionJitter

fn per_panel(&self) -> bool

ggplot2 4.0.3 measures position_jitter’s default width/height (resolution(x) * 0.4) from each facet panel’s own x/y, not the rows pooled across panels (#153) — verified with layer_data(): two panels with very different x spacing (e.g. integers vs. hundredths) get differently-scaled jitter even under fixed facet scales. The RNG draw is also per panel (it resets from seed at the top of compute, as it already did), matching ggplot2’s own per-panel restart — verified by the identical normalized jitter fractions ggplot2 produces across panels for the same seed.

fn compute(&self, data: &mut DataFrame, _params: &PositionParams)

Adjust positions for data.

fn name(&self) -> &str

fn group_order(&self, _data: &DataFrame) -> Option<Vec<String>>

The whole layer’s group rank order (first-appearance, pre-panel- split), for a per_panel position whose group ranking must come from the whole layer even though its per-x counting stays per-panel (review #167 finding 2; sipemu/ggplot-rs#153). build::compute_position_per_panel calls this on the pooled data before splitting and threads the result through PositionParams::group_order to every panel’s compute call. Default None — positions that don’t need this (stack/fill/ jitter) are unaffected.

fn validate(&self, _warnings: &mut Vec<String>)

Validate this position’s own parameters once per layer (D2); default no-op. Implementations push a warning and fall back to a sane default rather than panicking or silently miscomputing.