Column Types Reference

Methods

Every method on *migration.Table that declares a column, with the type each driver gets. The generic name in the first column is the ColType* constant, which you only need when calling AddColumn directly.

Integers

Method	PostgreSQL	MySQL	SQLite
`Increments(name)`	`SERIAL`	`INT`	`INTEGER`
`BigIncrements(name)`	`BIGSERIAL`	`BIGINT`	`INTEGER`
`TinyInt(name)`	`SMALLINT`	`TINYINT`	`TINYINT`
`SmallInt(name)`	`SMALLINT`	`SMALLINT`	`SMALLINT`
`MediumInt(name)`	`INTEGER`	`MEDIUMINT`	`MEDIUMINT`
`Int(name)`	`INTEGER`	`INT`	`INTEGER`
`BigInt(name)`	`BIGINT`	`BIGINT`	`BIGINT`

UnsignedTinyInt, UnsignedSmallInt, UnsignedMediumInt, UnsignedInt and UnsignedBigInt are the same types with Unsigned() applied — which adds UNSIGNED on MySQL and nothing on PostgreSQL or SQLite.

There is no Integer or BigInteger. The methods are Int and BigInt.

Decimals

Method	PostgreSQL	MySQL	SQLite
`Float(name, precision, scale)`	`FLOAT`	`FLOAT`	`FLOAT`
`Double(name, precision, scale)`	`DOUBLE PRECISION`	`DOUBLE`	`DOUBLE`
`Decimal(name, precision, scale)`	`DECIMAL`	`DECIMAL`	`DECIMAL`

Pass 0, 0 to Double unless you mean a precision — see the note under dates.

Text

Method	PostgreSQL	MySQL	SQLite
`Char(name, length)`	`CHAR`	`CHAR`	`CHAR`
`String(name, length)`	`VARCHAR`	`VARCHAR`	`VARCHAR`
`Text(name)`	`TEXT`	`TEXT`	`TEXT`
`Json(name)`	`JSONB`	`JSON`	`TEXT`
`UUID(name)`	`UUID`	`CHAR`	`TEXT`
`ULID(name)`	`CHAR`	`CHAR`	`TEXT`

Json gives JSONB on PostgreSQL rather than JSON: it stores a parsed form, so it is cheaper to read than to write and can carry a GIN index. The cost is that key order is not preserved and duplicate keys collapse — both properties of JSON objects rather than of this choice. SQLite has no JSON type; it stores JSON as TEXT and its JSON1 functions operate on TEXT, so that is the answer rather than a fallback.

MySQL cannot index a JSON column at all, and no key length helps. Validate says so, pointing at a generated column over the extracted value. The same check covers TEXT and BLOB columns, which MySQL refuses in a key without a length.

Dates and times

Method	PostgreSQL	MySQL	SQLite
`Date(name)`	`DATE`	`DATE`	`DATE`
`Time(name, precision)`	`TIME`	`TIME`	`TIME`
`DateTime(name, precision)`	`TIMESTAMP`	`DATETIME`	`DATETIME`
`DateTimeTz(name, precision)`	`TIMESTAMP WITH TIME ZONE`	`DATETIME`	`DATETIME`
`Timestamp(name, precision)`	`TIMESTAMP`	`TIMESTAMP`	`TIMESTAMP`
`TimestampTz(name, precision)`	`TIMESTAMP WITH TIME ZONE`	`TIMESTAMP`	`TIMESTAMP`

Pass 0 for the precision on the Tz variants. It is appended after the whole type name, so TimestampTz("created_at", 6) emits TIMESTAMP WITH TIME ZONE(6) and PostgreSQL answers syntax error at or near "(". With 0 you get the type and no suffix, which is what was wanted. DateTime and Timestamp take a precision safely, because theirs is one word.

Other

Method	PostgreSQL	MySQL	SQLite
`Boolean(name)`	`BOOLEAN`	`BOOLEAN`	`BOOLEAN`
`Binary(name)`	`BYTEA`	`BINARY`	`BINARY`
`Enum(name, values...)`	`TEXT` + `CHECK`	`ENUM(...)`	`TEXT`

Enum enforces nothing on SQLite. MySQL gets a native ENUM and PostgreSQL a TEXT with a CHECK (col IN (...)) appended, but SQLite gets plain TEXT and no constraint at all — so the values are checked on two drivers of three. Use a String and a Check when you need all three:

t.String("status", 16).Default("draft")
t.Check("posts_status_valid", "status IN ('draft','published','removed')")

Timestamps and soft deletes

There are no Timestamps() or SoftDeletes() shortcuts. Declare them:

t.TimestampTz("created_at", 0).Default(migration.Expr("now()"))
t.TimestampTz("updated_at", 0).Default(migration.Expr("now()"))
t.TimestampTz("deleted_at", 0).Nullable()

A soft-delete column must be nullable, and the model field must be *time.Time or sql.NullTime — the ORM rejects a bare time.Time at metadata time, because its zero value writes a real timestamp and deleted_at IS NULL would then match nothing and hide the whole table. See Soft deletes.

Schema Builder Writing Migrations