Site Selector

A dropdown in the top bar that lists every known measurement database, grouped by rig type, so visitors can jump straight to another site.

Overview

Where Cross-Site Search answers “who else measured this device”, the site selector answers “where else can I go”. It lists the databases themselves rather than individual devices, and highlights the one the visitor is currently on.

This works on any host — squig.link is not required. The selector reads a single small directory document (a few KB), so it costs one cached request per visit.

That document comes from GAA, which merges GraphAggregator’s federated index with squig.link’s site registry and publishes just the site and database list — no device corpus. modernGraphTool ships with its URL built in, so there is nothing to set up.

Getting your site listed

The selector shows what the shared index knows about. To appear in it:

• Self-hosted or custom domain — register at graphaggregator.harutohiroki.com.
• Hosted on squig.link — you are already listed, via squig.link’s own registry.

Registering also makes your databases discoverable through cross-site search on every other deployment, and it is what makes the selector appear on your own site under the default auto setting.

Configuration

Configured in config.js. The section is optional — omit it entirely and the defaults below apply:

SITE_SELECTOR: {
    ENABLED: 'auto',
    INDEX_URLS: [],
}

Configuration Options

• ENABLED: When to show the selector.
  • 'auto' (default) — show it only when this deployment is listed in the index, or is hosted on squig.link. A standalone site that isn’t registered gets nothing, rather than a dropdown listing a hundred other people’s databases and none of its own.
  • true — always show it.
  • false — never show it, and skip the fetch entirely.
• INDEX_URLS: Directory documents to try, in order. Leave empty to use the official GAA index.

Self-hosting the index

To avoid depending on a third-party origin, publish your own directory document and point INDEX_URLS at it:

SITE_SELECTOR: {
    ENABLED: 'auto',
    INDEX_URLS: [
        'https://example.com/my-site-index.json',
        'https://backup.example.com/my-site-index.json'
    ],
}

Each URL is tried in order and the first valid document wins, so a mirror can be listed as a backup. The document must follow the GAA schema — see the GAA repository for the spec and for a build script that generates it. Forking GAA and pointing this at your fork’s GitHub Pages URL is the least-effort route.

Notes

• The database the visitor is currently viewing is detected by URL and shown as selected. Longest path match wins, so a site hosting / and /headphones/ resolves to the right one.
• Databases that were unreachable when the index was last built stay listed but are dimmed, with a tooltip explaining that the link may not work. A site that is briefly down doesn’t vanish from navigation.
• Entries open in a new tab.