Datasets & filters

A dataset is data fetched into a keyed store element; bricks bound to that key render it live. Filters let a viewer reshape what they see.

Create a dataset

Add a data brick that writes to a target key:

• ApiData — fetch from an API connection or a public URL.
• SqlData — run a read-only query against a database.

Both support an optional refresh interval for live data.

Bind bricks to it

Set a brick's bindKey to the dataset's key. Charts, StatCards, Tables — even Text and Heading — read their value from that key and re-render when it changes. One dataset can feed many bricks.

Filters that re-query

1. Add an Input or Select and give it a bindKey, e.g. filter.region.
2. Reference that key in the dataset's params ({ from: "filter.region" }) and in the query (WHERE region = $1).
3. When the viewer changes the filter, the dataset re-queries and every bound brick updates.

This is the Superset/Metabase pattern — controls drive the query, the query drives the view.

Collections (repeaters)

To render a list of cards or rows, bind a Repeater to a dataset array; its one child is the per-record template, and child bricks read fields with bindField.

Next steps

• Mortar — transforms & computed values
• Build a dashboard