Performance

Every number in this section comes from a validated local protocol artifact captured on a clean tree; nothing is published from an unvalidated or protocol-mismatched artifact. Every expected cell carries either a validated timing or its recorded failure - a run that did not complete is shown as a failure, never converted into a timing.

Each page shows the results first, then the method and the command that reproduces them.

Sheetwrite 0.6.0 evidence

The render page includes chart-source tracking; the data-engine page includes incremental pivots. The release comparison shows fresh 0.5.4 and 0.6.0 measurements, recorded thresholds, every flagged row, and the conditions both sides were captured under. The original committed 0.5.x files remain under bench/results/releases/0.5.0/ as history only.

• Render performance: Scroll, edit, and interaction timings of the live grid, measured against Handsontable.
• Data engine performance: Ingest, window read, edit, sort, filter, and aggregate timings of the data store.
• Core write path: Timings of store and document writes at production sizes.
• Formula engine performance: Recalculation timings by dependency shape, and how the two engine builds compare.
• XLSX codec performance: Import and export time and peak memory for XLSX files.
• Delivery size: Published package sizes for each Sheetwrite release.
• Memory: Release retained engine memory, and read which owner holds each byte.
• Release comparison: Each release measured against the previous one on the same machine, suite by suite.
• Benchmark protocol: How the evidence is captured, and how to check a change for a timing regression.