Connections

Each dashboard’s connections.yaml declares the data sources it reads. PlotPress ships drivers for the most common SQL and document stores, plus DuckDB-backed access to S3.

Supported drivers

Driver	Best for	Page
files	CSV / JSON / Parquet files on disk — the simplest start	Files
duckdb	Local analytics over files + DuckDB databases	DuckDB
postgres	OLTP databases, warehouse replicas	PostgreSQL
mysql	MySQL 8.x	MySQL
mariadb	MariaDB 10.x+	MariaDB
clickhouse	Event analytics, wide tables	ClickHouse
mongodb	Document collections, app DBs	MongoDB
s3	Parquet / CSV in object storage (via DuckDB)	S3

The primary path: views

Every driver page documents the same default: a Plot block references a view (or its driver-specific equivalent) exposed by the database.

Plot.barY(data, { x: "month", y: "revenue" })

The dashboard team owns monthly_revenue in the database. The view’s name, columns, and types are the contract; the dashboard never re-derives them.

The fallback: queries/*.sql

When the database doesn’t expose a fitting view, drop a SQL file in the dashboard’s queries/ folder and reference it with query=. Treat it as a workaround — see Authoring → Queries as fallback.

Each driver page below has a “Fallback example” section showing the driver-specific shape (Postgres uses :name, ClickHouse uses {name:Type}, MongoDB uses an aggregation pipeline file, etc.).

Where each driver page goes

Every page follows the same structure:

1. DSN format — what to put in connections.yaml.
2. Authentication — where to source the password / token.
3. Views (primary) — what counts as a view in this driver.
4. Fallback (queries/) — driver-specific bind syntax, parameter handling.
5. Caveats — read-only enforcement, type quirks, pool tuning.