Data Grid Benchmark Plan

The first product-validation benchmark is a Vue data grid with a Vooya island implementation and a Vue-only baseline. It is not a synthetic Rust-versus-JS microbenchmark.

Workload

• 100,000 generated rows with name and numeric score columns.
• Virtual scrolling with a fixed visible window.
• Client-side text filtering and numeric sorting.
• Repeated filter and scroll interactions after initial load.

Measurements

• Compressed WASM and JavaScript asset sizes.
• Time from navigation to first interactive grid.
• Median and p95 interaction latency for filtering, sorting, and scrolling.
• Main-thread long tasks and peak memory under the same browser/device profile.
• The number of JS/WASM boundary crossings per interaction.

Decision rule

Continue Stage 4 only when the Vooya implementation improves at least one user-visible interaction under the target workload without introducing an unacceptable first-load regression. The report must present both wins and regressions; a faster isolated compute loop is insufficient evidence.

The first run and its decision are recorded in the Stage 3 result.