# hyper-p2p-histogram-gossip

Per-metric histograms with bucket gossip merge over Hyperswarm.

**Category:** measurement-rate-control

**Protocol:** `histogram-gossip/v1`

## When to use

Aggregating latency or resource samples across peers for approximate percentiles.

## When not to use

Exact global percentiles requiring centralized storage (use hyper-p2p-percentile-sketch locally).

## Quick start

```js
const { HyperP2PHistogramGossip } = require('hyper-p2p-histogram-gossip')
const m = new HyperP2PHistogramGossip({ topic: process.argv[2] })
await m.ready()
m.observe('latency', 42)
console.log(m.percentile('latency', 95))
await m.close()
```

## Docs

- [docs/api.md](docs/api.md)
- [docs/architecture.md](docs/architecture.md)

## Test

```bash
npm install && npm test
```