Self-hosted, Self-Improving, Federated Everything App — dev management, agent orchestration, persistent mind, digital twin, second brain, and creative pipeline from a single dashboard over Tailscale to port the user into the OS
JavaScript ai-agentsdashboarddeveloper-toolsdigital-twinquantified-selfsecond-brainself-hostedtailscale
/**
* Snapshot checksum helpers for cross-instance sync.
*
* A sync category's checksum is what lets the orchestrator skip the full
* payload transfer when nothing moved, so every snapshot getter needs one.
* These are the shared implementations — three getters had grown their own
* byte-identical copy (`dataSync.js`, `digital-twin-sync.js`, `peerUsage.js`).
*
* Two variants, and the choice matters:
*
* - `snapshotChecksum` hashes `JSON.stringify` output, so it is sensitive to
* key INSERTION ORDER. Correct only where the getter already canonicalizes
* ordering itself (sorted arrays, deterministically rebuilt objects).
* - `canonicalSnapshotChecksum` hashes `canonicalStringify`, which sorts object
* keys recursively — so two machines holding identical data hash the same
* regardless of the order they happened to write it in. Use this for any
* payload whose top level is a map keyed by something arriving over the wire.
*
* Getting that wrong is not a crash: it is two converged peers whose checksums
* never match, which the sync UI reads as "behind" forever.
*/
import crypto from 'crypto';
import { canonicalStringify } from './objects.js';
/** md5 of `JSON.stringify(data)` — insertion-order sensitive. */
export function snapshotChecksum(data) {
return crypto.createHash('md5').update(JSON.stringify(data)).digest('hex');
}
/** md5 of `canonicalStringify(data)` — stable across machines and write order. */
export function canonicalSnapshotChecksum(data) {
return crypto.createHash('md5').update(canonicalStringify(data)).digest('hex');
}