Platform Explorer Visualizer (egui)

Visual desktop app that pulls validator and block header data from Dash Platform Explorer’s HTTP API and visualizes the Gaussian-like distribution of proposed blocks per validator over selectable windows.

Features

• Timeframes: last 1 month, 3 months, 1 year.
• Uptime proxy filter: only validators that appear near both the start and end of the window (configurable grace days) are included.
• Histogram + normal fit: shows mean/std; overlays a scaled normal curve.
• Goodness-of-fit: chi-square statistic and p-value.
• Z-scores: top/bottom validators by z-score.
• Search: type a pro_tx_hash (full or prefix) to highlight its bar and show its count.
• Local persistence: caches data in a bundled SQLite DB (visualizer.db).

API

• Base (testnet / production): https://platform-explorer.pshenmic.dev
• Endpoints used:
  • GET /validators?page=&limit=
  • GET /blocks?page=&limit= (returns block header with timestamp and validator proTxHash)

Requirements

• Rust toolchain (Rust 1.77+ recommended).
• macOS, Linux, or Windows. Uses eframe/egui + OpenGL via glow.

Build & Run

• Build:
  • cd platform_explorer_visualizer
  • cargo build
• GUI app:
  • cargo run
  • Click “Sync data” to fetch (binary-searched to ~last year for speed), pick timeframe, adjust presence window, then “Recompute”.
  • Use the “Search pro_tx_hash” box to highlight and read the selected node’s count.
• Headless sync + quick summary (optional):
  • cargo run --bin sync

Data Storage

• SQLite database file visualizer.db is created in the current working directory.
• Tables:
  • validators(pro_tx_hash, is_active, proposed_blocks_amount)
  • blocks(height, ts_utc, validator)

Uptime Filter (Approximation)

• The API does not expose an explicit validator uptime over windows. We approximate by keeping validators that proposed at least one block within N days after the window start and at least one within N days before the window end (N is configurable; default 7 days).

Performance Notes

• Sync uses conservative paging limits and retry logic to avoid server 500s/timeouts.
• Initial block sync jumps directly to the page covering ~last year via binary search, then backfills new data only.
• Subsequent syncs are incremental (based on last stored block height).

Project Layout

• platform_explorer_visualizer/ crate with:
  • src/main.rs: egui app and plotting.
  • src/syncer.rs: HTTP fetching, pagination, binary search.
  • src/store.rs: SQLite schema and queries; windowed counts with presence filter.
  • src/types.rs: API JSON models.
  • src/bin/sync.rs: optional headless sync runner.

Troubleshooting

• If sync shows intermittent errors, re-run “Sync data”; the client retries and uses smaller pages.
• If the histogram is empty, ensure you clicked “Recompute” after syncing and the presence window isn’t too strict.
• On systems without proper OpenGL drivers, consider running the headless sync only.

Roadmap Ideas

• Parameterize sync cutoff by selected timeframe (only fetch needed pages).
• Replace presence proxy with explicit membership/uptime if/when the API exposes it.
• Export CSV and add per-validator detail view.