Struct SelectionVector 

pub struct SelectionVector { /* private fields */ }

A selection vector tracks qualifying row indices

This is the core data structure for late materialization. Instead of copying row data, we track which row indices qualify and only materialize when needed.

Memory Layout

Uses u32 indices (4 bytes each) instead of full rows (potentially hundreds of bytes each). For a 1M row table filtered to 10K rows:

• Old: 1M × sizeof(Row) = potentially gigabytes
• New: 10K × 4 bytes = 40KB

Example

// Create from filter bitmap
let selection = SelectionVector::from_bitmap(&[true, false, true, true, false]);
assert_eq!(selection.len(), 3);
assert_eq!(selection.indices(), &[0, 2, 3]);

// Iterate over qualifying indices
for idx in selection.iter() {
    let row = &source_rows[idx as usize];
    // Process row...
}

Implementations

impl SelectionVector

pub fn empty() -> Self

Create an empty selection vector

pub fn all(count: usize) -> Self

Create a selection vector that selects all rows in range [0, count)

pub fn from_range(range: Range<usize>) -> Self

Create a selection vector from a range

pub fn from_indices(indices: Vec<u32>) -> Self

Create from pre-computed indices (assumes sorted, unique)

pub fn from_bitmap(bitmap: &[bool]) -> Self

Create from a boolean filter bitmap

This is the most common creation path after evaluating predicates.

Example

let bitmap = vec![true, false, true, true, false];
let selection = SelectionVector::from_bitmap(&bitmap);
assert_eq!(selection.indices(), &[0, 2, 3]);

pub fn from_bitmap_with_capacity(bitmap: &[bool], capacity_hint: usize) -> Self

Create from a bitmap with pre-allocated capacity hint

Use when you have an estimate of how many rows will match.

pub fn len(&self) -> usize

Number of selected rows

pub fn is_empty(&self) -> bool

Check if no rows are selected

pub fn indices(&self) -> &[u32]

Get the underlying indices slice

pub fn into_indices(self) -> Vec<u32>

Consume and return the indices vector

pub fn get(&self, pos: usize) -> Option<u32>

Get index at position

pub fn iter(&self) -> impl Iterator<Item = u32> + '_

Iterate over selected indices

pub fn intersect(&self, other: &SelectionVector) -> SelectionVector

Intersect with another selection vector (AND operation)

Returns indices that appear in both selections. Useful for combining multiple filter conditions.

pub fn union(&self, other: &SelectionVector) -> SelectionVector

Union with another selection vector (OR operation)

Returns indices that appear in either selection.

pub fn remap(&self, base: &SelectionVector) -> SelectionVector

Remap indices through another selection

If self contains indices [0, 2, 3] and base contains [10, 20, 30, 40, 50], returns [10, 30, 40] (indices 0, 2, 3 from base).

This is used when chaining operations: if we filter a filtered result, we need to map back to original row indices.

pub fn to_bitmap(&self, total_count: usize) -> Vec<bool>

Create a dense bitmap representation

Returns a boolean vector where result[i] = true iff i is in the selection. total_count is the total number of rows in the source.

pub fn filter<T, F>(&self, data: &[T], predicate: F) -> SelectionVector

where F: Fn(&T) -> bool,

Filter this selection based on a predicate applied to a slice of data

This is useful for applying additional filters without materializing rows.

pub fn map<T, F>(&self, f: F) -> Vec<T>

where F: FnMut(u32) -> T,

Apply a function to each selected index, collecting results

pub fn selectivity(&self, total_count: usize) -> f64

Selectivity ratio (selected / total)

Returns a value between 0.0 and 1.0 indicating what fraction of the total rows are selected.

pub fn is_contiguous(&self) -> bool

Compact representation check

Returns true if this selection represents a contiguous range, which allows for more efficient processing.

pub fn as_range(&self) -> Option<Range<usize>>

Get the range if this selection is contiguous

Trait Implementations

impl Clone for SelectionVector

fn clone(&self) -> SelectionVector

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for SelectionVector

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

Formats the value using the given formatter.

impl Default for SelectionVector

fn default() -> Self

Returns the “default value” for a type.

impl From<Vec<u32>> for SelectionVector

fn from(indices: Vec<u32>) -> Self

Converts to this type from the input type.

impl From<Vec<usize>> for SelectionVector

fn from(indices: Vec<usize>) -> Self

Converts to this type from the input type.

impl<'a> IntoIterator for &'a SelectionVector

type Item = &'a u32

The type of the elements being iterated over.

type IntoIter = Iter<'a, u32>

Which kind of iterator are we turning this into?

fn into_iter(self) -> Self::IntoIter

Creates an iterator from a value.

impl IntoIterator for SelectionVector

type Item = u32

The type of the elements being iterated over.

type IntoIter = IntoIter<u32>

Which kind of iterator are we turning this into?

fn into_iter(self) -> Self::IntoIter

Creates an iterator from a value.