Low-Level Types

Low-level types give you direct control over SQL column types when semantic types don’t fit your needs.

Type Reference

Type	SQL Result	Notes
`col.string(n)`	VARCHAR(n)	Fixed-length string
`col.text()`	TEXT	Unlimited length
`col.integer()`	INTEGER	32-bit integer
`col.float()`	FLOAT	Floating point
`col.decimal(p,s)`	DECIMAL	Precise decimal
`col.boolean()`	BOOLEAN	True/false
`col.date()`	DATE	Date only
`col.time()`	TIME	Time only
`col.datetime()`	TIMESTAMP	Date and time
`col.uuid()`	UUID	Unique identifier
`col.json()`	JSONB	JSON data
`col.base64()`	TEXT	Base64 encoded
`col.enum([...])`	CHECK	Constrained values

Examples

Strings

export default table({
  id: col.id(),
  sku: col.string(50).unique(), // VARCHAR(50)
  description: col.text().optional(), // TEXT, nullable
});

Numbers

export default table({
  id: col.id(),
  amount: col.decimal(10, 2), // DECIMAL(10,2)
  count: col.integer(), // INTEGER
  rate: col.float(), // FLOAT
});

Dates and Times

export default table({
  id: col.id(),
  event_date: col.date(), // DATE
  start_time: col.time(), // TIME
  created_at: col.datetime(), // TIMESTAMP
});

JSON and UUID

export default table({
  id: col.uuid(), // UUID (use col.id() for primary key)
  user: col.belongs_to("auth.user"),
  preferences: col.json(), // JSONB
});

Enums

export default table({
  id: col.id(),
  status: col
    .enum(["pending", "processing", "shipped", "delivered"])
    .default("pending"),
  priority: col.enum(["low", "normal", "high"]).default("normal"),
}).timestamps();

Prefer Semantic Types

For most cases, use semantic types instead - they include validation and sensible defaults:

Instead of…	Use…
`col.string(255)`	`col.email()`
`col.string(100)`	`col.name()`
`col.integer()`	`col.quantity()`
`col.decimal(19,4)`	`col.money()`
`col.boolean()`	`col.flag()`

See Semantic Types for the full list.