Type mapping

Each language guide below shows how SQL types map to host-language types, how nullability is expressed, and which dependencies are required.

Nullability

Nullability is determined from the schema and the query structure:

• A column with NOT NULL is non-null.
• A column without NOT NULL is nullable.
• PRIMARY KEY columns are implicitly non-null.
• Columns from the outer side of a LEFT/RIGHT/FULL JOIN are made nullable.
• Scalar subqueries in the SELECT list produce nullable results.

The sections below show the nullable form alongside the non-null form for each type.

Java

Reference types (String, BigDecimal, etc.) are nullable by passing null. Primitive types (boolean, int, long, etc.) are boxed when the column is nullable.

The generated code uses only java.sql and standard JDK classes. No extra imports are needed for basic types.

Kotlin

Rust

Enable sqlx features as needed:

sqlx = { version = "0.8", features = [
    "runtime-tokio",
    "postgres",      # or "sqlite" / "mysql"
    "time",          # Date, Time, Timestamp, TimestampTz
    "uuid",          # UUID
    "rust_decimal",  # NUMERIC / DECIMAL
] }

Python

All date/time, decimal, and uuid types come from the standard library. psycopg3 automatically deserializes JSON columns to Python objects; sqlite3 and mysql-connector return the raw JSON string.

TypeScript

⚠️ BIGINT maps to number, which loses precision above 2⁵³. Use BigInt in application code if your IDs or values may exceed this range.

All types are built-in to TypeScript/Node.js — no extra dependencies needed for the type annotations themselves.

JavaScript

Type mapping is identical to TypeScript. Types are expressed as JSDoc comments (@typedef, @param, @returns) rather than inline TypeScript syntax, but the underlying types are the same.

Go

All nullable types use the standard database/sql null wrappers. No extra dependencies are required beyond the driver itself.