Struct Compiled 

pub struct Compiled {
    pub domain: Domain,
    pub problem: Problem,
    pub minimize: bool,
    pub n_prefs: usize,
    pub warn_other: bool,
    pub unsupported: Option<String>,
    pub synthetic: HashSet<String>,
    pub forgos: Vec<(String, f64)>,
    pub folded_metric: bool,
}

Fields

domain: Domain

problem: Problem

minimize: bool

n_prefs: usize

warn_other: bool

unsupported: Option<String>

Set if the metric is outside the supported class (maximize / negative weight / scaled or non-monotone total-cost). The caller falls back to a satisficing plan instead of silently optimizing the wrong objective.

synthetic: HashSet<String>

Names of the synthetic Keyder-Geffner actions (stripped from the plan).

forgos: Vec<(String, f64)>

(forgo-action name, weight) per preference instance.

folded_metric: bool

True when a numeric metric term was FOLDED into total-cost (mirrored increase effects on real actions, e.g. rovers’ traverse costs). The metric optimizer routes such tasks to the legacy compiled-goal B&B — real-action cost gives it a genuine gradient, and the closure search measures worse there (continuous tightening churn).