Struct egg::BackoffScheduler

pub struct BackoffScheduler { /* private fields */ }

A RewriteScheduler that implements exponentional rule backoff.

For each rewrite, there exists a configurable initial match limit. If a rewrite search yield more than this limit, then we ban this rule for number of iterations, double its limit, and double the time it will be banned next time.

This seems effective at preventing explosive rules like associativity from taking an unfair amount of resources.

BackoffScheduler is configurable in the builder-pattern style.

Implementations

impl BackoffScheduler

pub fn with_initial_match_limit(self, limit: usize) -> Self

Set the initial match limit after which a rule will be banned. Default: 1,000

pub fn with_ban_length(self, ban_length: usize) -> Self

Set the initial ban length. Default: 5 iterations

pub fn do_not_ban(self, name: impl Into<Symbol>) -> Self

Never ban a particular rule.

pub fn rule_match_limit(self, name: impl Into<Symbol>, limit: usize) -> Self

Set the initial match limit for a rule.

pub fn rule_ban_length(self, name: impl Into<Symbol>, length: usize) -> Self

Set the initial ban length for a rule.

Trait Implementations

impl Debug for BackoffScheduler

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

Formats the value using the given formatter.

impl Default for BackoffScheduler

fn default() -> Self

Returns the “default value” for a type.

impl<L, N> RewriteScheduler<L, N> for BackoffSchedulerwhere L: Language, N: Analysis<L>,

fn can_stop(&mut self, iteration: usize) -> bool

Whether or not the Runner is allowed to say it has saturated.

fn search_rewrite<'a>( &mut self, iteration: usize, egraph: &EGraph<L, N>, rewrite: &'a Rewrite<L, N> ) -> Vec<SearchMatches<'a, L>>

A hook allowing you to customize rewrite searching behavior. Useful to implement rule management.

fn apply_rewrite( &mut self, iteration: usize, egraph: &mut EGraph<L, N>, rewrite: &Rewrite<L, N>, matches: Vec<SearchMatches<'_, L>> ) -> usize

A hook allowing you to customize rewrite application behavior. Useful to implement rule management.