DDL — Data Definition

CREATE TABLE

CREATE TABLE table_name (
    column1 TYPE,
    column2 TYPE,
    ...
);

Supported Types

Type	Description	Example
TEXT	UTF-8 string	'hello'
INTEGER	64-bit signed integer	42
REAL	64-bit floating point	3.14
BOOL	Boolean	TRUE, FALSE
BLOB	Binary data	X'DEADBEEF'

Examples

-- Basic table
CREATE TABLE users (
    id TEXT,
    name TEXT,
    email TEXT,
    age INTEGER
);

-- Create only if it doesn't exist
CREATE TABLE IF NOT EXISTS events (
    id TEXT,
    type TEXT,
    payload TEXT,
    timestamp INTEGER
);

Note

TensorDB tables don’t have traditional PRIMARY KEY constraints. The first column is conventionally used as the row identifier for key encoding (table/<name>/<pk_value>).

DROP TABLE

DROP TABLE table_name;
DROP TABLE IF EXISTS table_name;

Dropping a table removes its metadata from __meta/table/<name>. Row data becomes orphaned but doesn’t affect the ledger’s integrity.

ALTER TABLE

ADD COLUMN

ALTER TABLE users ADD COLUMN phone TEXT;

Adds a new column to the table schema. Existing rows return NULL for the new column.

DROP COLUMN

ALTER TABLE users DROP COLUMN phone;

Removes a column from the table schema. This is a metadata-only operation — existing rows are not rewritten. The dropped column is excluded from all future projections. Primary key columns cannot be dropped.

RENAME COLUMN

ALTER TABLE users RENAME COLUMN email TO email_address;

Renames a column in the table schema. A column_aliases mapping is maintained so that existing rows stored with the old column name are automatically mapped to the new name on read.

Caution

ALTER TABLE operations modify table metadata only. Existing rows on disk are not rewritten — ADD COLUMN returns NULL on old rows, DROP COLUMN excludes the column from projections, and RENAME COLUMN uses an alias map for backward compatibility.