Module factorized_iter 

Improved iteration over factorized data.

This module provides cache-friendly iteration over FactorizedChunk data using pre-computed indices for better performance on multi-hop traversals.

Performance

The original FactorizedRowIterator computes ranges on-the-fly, which can be cache-unfriendly for deep traversals. This improved iterator:

• Pre-computes all indices before iteration starts
• Uses SmallVec for stack allocation when level count is small
• Provides RowView for zero-copy value access

Example

let chunk = get_factorized_chunk();

// Pre-computed iteration (better for repeated access)
let iter = PrecomputedIter::new(&chunk);
for row in iter {
    let source_id = row.get_node_id(0, 0);
    let dest_id = row.get_node_id(1, 0);
    // ... process row
}

Structs

PrecomputedIter

Pre-computed iterator over logical rows in a factorized chunk.

RowIndices

Pre-computed indices for a single logical row.

RowView

A view into a single logical row of a factorized chunk.

StreamingIter

Streaming iterator that doesn’t pre-compute indices.