Skip to content
Tabular data visualization engine

Roadmap

Current State

Section titled “Current State”

Vizb is a data-visualization tool. It turns CSV and JSON tables into interactive 4D HTML charts. It also ships parsers that normalize benchmark output from three languages across five frameworks:

  • Go — go test -bench output (raw text and -json)
  • Rust — cargo bench via Criterion (rs:criterion) and Divan (rs:divan)
  • JavaScript / TypeScript — Vitest bench (js:vitest) and Tinybench console.table (js:tinybench)

Every parser — CSV, JSON, and benchmark — supports multi-dimensional grouping, tag-based merging, time unit conversion, filtering, and auto-aggregation of grouped rows. See the Supported Inputs guide for setup.

All inputs visualize the same way, up to four dimensions. Add a z-axis for interactive 3D charts with auto-rotate and axis-sum tooltips. Or use a flat heatmap. Charts ship as bar, line, scatter, pie, heatmap, and radar. Each has a statistics panel with descriptive metrics and a four-method correlation matrix (Pearson, Spearman, Kendall, distance correlation).

Planned

Section titled “Planned”

Vizb currently focuses on linear, tabular data. The work below extends supported inputs and visual shapes.

More Inputs

Section titled “More Inputs”

New benchmark adapters normalize another tool’s output into the same table vizb already charts.

Python

Integrate with pytest-benchmark output. Generate interactive visualizations for Python performance testing.

C++

Parse output from Google Benchmark and similar C++ benchmarking frameworks. Bring multi-dimensional visualization to native performance testing.

Dart

Support the Dart benchmark harness (benchmark_harness package). Visualize Dart VM performance data.

Java

Parse output from JMH (Java Microbenchmark Harness) and similar Java benchmarking frameworks. Bring multi-dimensional visualization to JVM performance testing.

Non-Linear Visualizations

Section titled “Non-Linear Visualizations”

We are exploring visualization types for data that is not well served by linear x/y/z charts:

  • Network / graph views for relationship data.
  • Tree and hierarchy layouts for nested or path-like data.
  • Geospatial maps for location-based datasets.

These are research areas. They will land after the input-parser ecosystem is stable.

Contributing

Section titled “Contributing”

Interested in helping? Check out the GitHub repository for open issues and contribution guidelines.