Enum Stage 

pub enum Stage {
    Scan {
        table: String,
        binding: String,
        rows: usize,
    },
    Join {
        kind: JoinKind,
        table: String,
        binding: String,
        strategy: Strategy,
        keys: usize,
        left_rows: usize,
        right_rows: usize,
        out_rows: usize,
        early_stopped: bool,
        post_filter_removed: Option<usize>,
    },
    Filter {
        in_rows: usize,
        out_rows: usize,
    },
    Project {
        columns: usize,
        out_rows: usize,
    },
    Distinct {
        in_rows: usize,
        out_rows: usize,
    },
    Sort {
        keys: usize,
        rows: usize,
    },
    Prefilter {
        binding: String,
        predicates: usize,
        in_rows: usize,
        out_rows: usize,
    },
    Limit {
        limit: Option<usize>,
        offset: Option<usize>,
        in_rows: usize,
        out_rows: usize,
    },
}

One stage of the pipeline that ran.

Variants

Scan

A base relation was materialised.

Fields

table: String

binding: String

The name its columns are addressed by — the alias when one was given. Shown because a plan that does not name its bindings is unreadable for a self-join.

rows: usize

Join

Two inputs were joined.

Fields

kind: JoinKind

table: String

binding: String

strategy: Strategy

keys: usize

Equality key pairs the planner could PROVE usable. Zero means the hash path was unavailable, which is the single most useful number in the plan when a join is unexpectedly slow.

left_rows: usize

right_rows: usize

out_rows: usize

early_stopped: bool

The join stopped early because the row budget was already met.

post_filter_removed: Option<usize>

Rows this join produced and then discarded because the WHERE clause was evaluated inside it. None means the filter ran as a separate stage.

Reported separately from the join’s own row count precisely so the two stay distinguishable: an ON predicate and a post-join WHERE predicate mean different things, and the plan should not blur them just because one loop evaluates both.

Filter

A WHERE clause was applied.

Fields

in_rows: usize

out_rows: usize

Project

The select list was evaluated.

Fields

columns: usize

out_rows: usize

Distinct

DISTINCT removed duplicates.

Fields

in_rows: usize

out_rows: usize

Sort

ORDER BY sorted the rows.

Fields

keys: usize

rows: usize

Prefilter

A WHERE conjunct was pre-applied to a base relation before the join.

The original WHERE still runs afterwards — this is a copy, not a move, which is what makes it safe for every join type.

Fields

binding: String

predicates: usize

in_rows: usize

out_rows: usize

Limit

LIMIT / OFFSET were applied.

Fields

limit: Option<usize>

offset: Option<usize>

in_rows: usize

out_rows: usize

Trait Implementations

impl Clone for Stage

fn clone(&self) -> Stage

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for Stage

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

Formats the value using the given formatter.

impl Eq for Stage

impl PartialEq for Stage

fn eq(&self, other: &Stage) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for Stage