Struct rustc_data_structures::transitive_relation::TransitiveRelation

pub struct TransitiveRelation<T> {
    builder: Frozen<TransitiveRelationBuilder<T>>,
    closure: Frozen<BitMatrix<usize, usize>>,
}

Fields

builder: Frozen<TransitiveRelationBuilder<T>>

closure: Frozen<BitMatrix<usize, usize>>

Implementations

impl<T: Eq + Hash + Copy> TransitiveRelation<T>

pub fn maybe_map<F, U>(&self, f: F) -> Option<TransitiveRelation<U>>where
    F: FnMut(T) -> Option<U>,
    U: Clone + Debug + Eq + Hash + Copy,

Applies the (partial) function to each edge and returns a new relation including transitive closures.

pub fn contains(&self, a: T, b: T) -> bool

Checks whether a < target (transitively)

pub fn reachable_from(&self, a: T) -> Vec<T>

Thinking of x R y as an edge x -> y in a graph, this returns all things reachable from a.

Really this probably ought to be impl Iterator<Item = &T>, but I’m too lazy to make that work, and – given the caching strategy – it’d be a touch tricky anyhow.

pub fn postdom_upper_bound(&self, a: T, b: T) -> Option<T>

Picks what I am referring to as the “postdominating” upper-bound for a and b. This is usually the least upper bound, but in cases where there is no single least upper bound, it is the “mutual immediate postdominator”, if you imagine a graph where a < b means a -> b.

This function is needed because region inference currently requires that we produce a single “UB”, and there is no best choice for the LUB. Rather than pick arbitrarily, I pick a less good, but predictable choice. This should help ensure that region inference yields predictable results (though it itself is not fully sufficient).

Examples are probably clearer than any prose I could write (there are corresponding tests below, btw). In each case, the query is postdom_upper_bound(a, b):

// Returns Some(x), which is also LUB.
a -> a1 -> x
           ^
           |
b -> b1 ---+

// Returns `Some(x)`, which is not LUB (there is none)
// diagonal edges run left-to-right.
a -> a1 -> x
  \/       ^
  /\       |
b -> b1 ---+

// Returns `None`.
a -> a1
b -> b1

pub fn mutual_immediate_postdominator<'a>(&'a self, mubs: Vec<T>) -> Option<T>

Viewing the relation as a graph, computes the “mutual immediate postdominator” of a set of points (if one exists). See postdom_upper_bound for details.

pub fn minimal_upper_bounds(&self, a: T, b: T) -> Vec<T>

Returns the set of bounds X such that:

• a < X and b < X
• there is no Y != X such that a < Y and Y < X
  • except for the case where X < a (i.e., a strongly connected component in the graph). In that case, the smallest representative of the SCC is returned (as determined by the internal indices).

Note that this set can, in principle, have any size.

pub fn parents(&self, a: T) -> Vec<T>

Given an element A, returns the maximal set {B} of elements B such that

• A != B
• A R B is true
• for each i, j: B[i] R B[j] does not hold

The intuition is that this moves “one step up” through a lattice (where the relation is encoding the <= relation for the lattice). So e.g., if the relation is -> and we have

a -> b -> d -> f
|              ^
+--> c -> e ---+

then parents(a) returns [b, c]. The postdom_parent function would further reduce this to just f.

fn with_closure<OP, R>(&self, op: OP) -> Rwhere
    OP: FnOnce(&BitMatrix<usize, usize>) -> R,

pub fn base_edges(&self) -> impl Iterator<Item = (T, T)> + '_

Lists all the base edges in the graph: the initial non-transitive set of element relations, which will be later used as the basis for the transitive closure computation.

Trait Implementations

impl<T: Clone> Clone for TransitiveRelation<T>

fn clone(&self) -> Self

Returns a copy of the value.

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

Performs copy-assignment from source.

impl<T: Debug> Debug for TransitiveRelation<T>

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

Formats the value using the given formatter.

impl<T> Deref for TransitiveRelation<T>

type Target = Frozen<TransitiveRelationBuilder<T>>

The resulting type after dereferencing.

fn deref(&self) -> &Self::Target

Dereferences the value.

Layout

Note: Most layout information is completely unstable and may even differ between compilations. The only exception is types with certain repr(...) attributes. Please see the Rust Reference’s “Type Layout” chapter for details on type layout guarantees.

Size: 120 bytes