Skip to main content

RepositionStrategy

elevator_core::dispatch

Trait RepositionStrategy 

Source 
pub trait RepositionStrategy: Send + Sync {
    // Required method
    fn reposition(
        &mut self,
        idle_elevators: &[(EntityId, f64)],
        stop_positions: &[(EntityId, f64)],
        group: &ElevatorGroup,
        world: &World,
        out: &mut Vec<(EntityId, EntityId)>,
    );

    // Provided methods
    fn builtin_id(&self) -> Option<BuiltinReposition> { ... }
    fn min_arrival_log_window(&self) -> u64 { ... }
}
Expand description

Pluggable strategy for repositioning idle elevators.

After the dispatch phase, elevators that remain idle (no pending assignments) are candidates for repositioning. The strategy decides where each idle elevator should move to improve coverage and reduce expected response times.

Implementations receive the set of idle elevator positions and the group’s stop positions, then return a target stop for each elevator (or None to leave it in place).

Required Methods§

Source

fn reposition( &mut self, idle_elevators: &[(EntityId, f64)], stop_positions: &[(EntityId, f64)], group: &ElevatorGroup, world: &World, out: &mut Vec<(EntityId, EntityId)>, )

Decide where to reposition idle elevators.

Push (elevator_entity, target_stop_entity) pairs into out. The buffer is cleared before each call — implementations should only push, never read prior contents. Elevators not pushed remain idle.

Provided Methods§

Source

fn builtin_id(&self) -> Option<BuiltinReposition>

If this strategy is a known built-in variant, return it so Simulation::set_reposition callers don’t have to pass a separate BuiltinReposition id that might drift from the dispatcher’s actual type.

Mirrors the pattern introduced for DispatchStrategy::builtin_id in #410: the runtime impl identifies itself so the snapshot identity always matches the executing behaviour, instead of depending on the caller to keep two parameters consistent. Default None — custom strategies should override to return BuiltinReposition::Custom with a stable name for snapshot fidelity.

Source

fn min_arrival_log_window(&self) -> u64

Minimum ArrivalLog retention (in ticks) the strategy needs to function. Strategies that read the log directly with a custom rolling window must override this so Simulation::set_reposition can widen ArrivalLogRetention to keep the data alive long enough for the query.

Default 0 — strategies that don’t read the arrival log (or that only consume it through DispatchManifest::arrivals_at, which already tracks retention) impose no requirement.

Implementors§