# Data types

> The column types of NYXDB, their spellings, bounds and index capabilities, and the NULL rule.

The table below lists the column types and the spellings `CREATE TABLE` accepts.
Type names are case-insensitive.

## Column types

| Spelling | Values | Bounds | Index structures |
| --- | --- | --- | --- |
| `BOOL`, `BOOLEAN` | `TRUE`, `FALSE` | | `BLOOM`, `POSTINGS` |
| `INT8`, `INT16`, `INT32`, `INT64` | signed integers of 8 to 64 bits | the type's range | `MINMAX`, `BLOOM`, `POSTINGS` |
| `INT128`, `INT256` | signed integers of 128 and 256 bits | the type's range | `MINMAX`, `BLOOM`, `POSTINGS` |
| `UINT8`, `UINT16`, `UINT32`, `UINT64` | unsigned integers of 8 to 64 bits | the type's range | `MINMAX`, `BLOOM`, `POSTINGS` |
| `UINT128`, `UINT256` | unsigned integers of 128 and 256 bits | the type's range | `MINMAX`, `BLOOM`, `POSTINGS` |
| `FLOAT32`, `FLOAT64` | IEEE binary floating point | | `MINMAX` |
| `DECIMAL(p, s)` | exact decimal with precision `p` and scale `s` | 1 to 76 digits of precision; `s` at most `p` | `MINMAX`, `BLOOM`, `POSTINGS` |
| `STRING` | UTF-8 text, checked on write | at most 16 MiB per value | `MINMAX`, `BLOOM`, `POSTINGS` |
| `BYTES` | a byte string | at most 16 MiB per value | `MINMAX`, `BLOOM`, `POSTINGS` |
| `FIXEDSTRING(n)` | a byte string of fixed length `n` | `n` from 1 to 1,048,576 | `MINMAX`, `BLOOM`, `POSTINGS` |
| `DATE` | a calendar date | years 0001 to 9999 | `MINMAX`, `BLOOM`, `POSTINGS` |
| `DATETIME64(p)`, `DATETIME64(p, 'tz')` | a UTC instant with `p` fractional-second digits | `p` from 0 to 9; the time-zone text at most 64 bytes; years 0001 to 9999 | `MINMAX`, `BLOOM`, `POSTINGS` |
| `UUID` | a 128-bit identifier | | `MINMAX`, `BLOOM`, `POSTINGS` |

Parameterized types are checked against their bounds when the statement is
parsed. `VALID BY` and `REVISION BY` name integer or timestamp columns. `ENCODING LOW_CARDINALITY` is dictionary encoding and is intended for
`STRING` columns with few distinct values.

## Writing values

Values are written as SQL literals. `BOOL` takes `TRUE` or `FALSE`; numeric
types take a number; `DECIMAL` takes a number or a string. `STRING`, `BYTES`,
`FIXEDSTRING`, `DATE` (`'YYYY-MM-DD'`), `DATETIME64`
(`'YYYY-MM-DDTHH:MM:SS[.fraction]Z'`) and `UUID` take a string. The full list is
on [INSERT](/docs/sql/insert).

## NULL

- A column is `NOT NULL` unless `NULL` is written after its type in
  `CREATE TABLE` or `ADD COLUMN`.
- A write that omits a column uses its `DEFAULT`, or NULL when the column is
  nullable.
- Primary-key columns and the columns named by `VALID BY` and `REVISION BY` are
  never null.
- A comparison with a NULL operand is NULL, `AND`, `OR` and `NOT` follow SQL
  three-valued logic, and a `WHERE` keeps only rows whose condition is true.
  Test for nulls with `IS NULL` and `IS NOT NULL`.

## Composite and network types

| Spelling | Values |
| --- | --- |
| `ARRAY(T)` | an ordered list of values of type `T` |
| `TUPLE(T1, T2, …)` | a fixed sequence of typed fields |
| `MAP(K, V)` | key–value pairs |
| `JSON` | a JSON document |
| `IPV4`, `IPV6` | network addresses |
| `ENUM` | one of a declared set of names |

Nullability is the `NULL` column option, and dictionary encoding is
`ENCODING LOW_CARDINALITY`.

---

Source: https://nyxdb.io/docs/sql/data-types/