Query recipes

Use these focused paths when you already understand installation and snapshot generation. Examples are labeled by scope so an inline proof is not mistaken for a runnable application.

Level	Scope	Verification
Inline proof	A small query or API boundary inside a focused guide	Backed by a package or documentation contract when it claims exact output
Recipe	One task with schema assumptions, compiler contract, rendered driver input, dialect notes, and source	Compiled fixture plus focused runtime assertions
Runnable application	Schema, config, generated evidence, adapter lifecycle, and broader capabilities	Service-free tests and a real-database suite

Construct queries

Task	Example	What it proves
Select with typed parameters	Your first query	Selected row fields and ordered parameters
Add optional filters or columns	Conditional composition	Explicit structural branches and result shape
Insert several rows from Array.map	Multi-row insert recipe	Automatic comma separation, compiler evidence, and five rows of ordered parameters
Choose mapped arrays or sql.join()	Explicit list controls	Empty, dynamic, or custom-separator behavior
Escape static analysis deliberately	Dynamic SQL	Explicit Query<unknown> boundary

Execute queries

Task	Example	What it proves
Execute and assert cardinality	Execution	all, one, maybeOne, cancellation, and deadlines
Use an existing pool	Existing pools	Application lifecycle remains application-owned
Prepare repeated query shapes	Prepared queries	Stable skeleton and bounded cardinality variants
Run an ordered batch	Batches	One lease, ordered results, and non-atomic default
Stream a large result	Streaming	Backpressure and explicit cleanup
Validate decoded rows	Result validation	Standard Schema checks after driver decoding

Move data and operate safely

Task	Example	What it proves
Choose multi-row SQL, batch, or native bulk	Bulk data	Different semantics and protocol capabilities
Trace query lifecycle	Observability	Neutral events, redaction, and OpenTelemetry
Route reads and retry transactions	Routing and retries	Semantic routing and bounded retry policy
Combine PostgreSQL and SQLite	Multi-database application	Separate grammar, config, evidence, and driver boundaries

Complete applications

For setup, generated evidence, runtime behavior, and database-backed tests in one place, use the PostgreSQL, MySQL, SQLite, or PostgreSQL plus SQLite application.

The source applications are exercised by the repository's example and real-database test gates. The pages link to their exact source files rather than presenting generated schema output as an application-facing API.