Struct RuleSetBuilder 

pub struct RuleSetBuilder<'outer> { /* private fields */ }

Builder for a RuleSet.

There are in general two ways to add rules to a rule set:

1. Use the QueryBuilder and RuleBuilder APIs to construct a rule from scratch.
2. Use a previously cached plan and add extra constraints to it.

The pattern this is used by egglog is as follows: An egglog rule is first compiled to a cached plan using builder patterns at declaration time, and each time the rule is run, it is added to a ruleset using the cached plan and possibly some extra constraints (e.g., timestamp).

See Database::new_rule_set for more information.

Implementations

impl<'outer> RuleSetBuilder<'outer>

pub fn new(db: &'outer mut Database) -> Self

pub fn estimate_size(&self, table: TableId, c: Option<Constraint>) -> usize

Estimate the size of the subset of the table matching the given constraint.

This is a wrapper around the Database::estimate_size method.

pub fn new_rule<'a>(&'a mut self) -> QueryBuilder<'outer, 'a>

Add a rule to this rule set.

pub fn add_rule_from_cached_plan( &mut self, cached: &CachedPlan, extra_constraints: &[(AtomId, Constraint)], ) -> Option<RuleId>

Add a rule to this rule set based on a previously cached plan, optionally with additional constraints applied on top.

Returns None if the query is provably empty given the current database state (i.e. some constraint narrows a table to zero matching rows), in which case no rule or action is allocated. Returns Some(RuleId) otherwise.

The primary use-case is seminaive evaluation: an egglog rule is compiled once into a CachedPlan and then added to a fresh RuleSet each iteration with timestamp constraints (e.g. GeConst on the focus atom) that select only new tuples. If no new tuples exist for an atom, the None return allows the caller to skip that variant entirely.

pub fn build(self) -> RuleSet

Build the ruleset.