SQLite¶

Scythe supports SQLite with its simplified type affinity system across all 10 languages. SQLite support operates at the parser and analyzer level -- SQL parsing, type inference, and nullability analysis are fully SQLite-aware. The code generation backends work the same regardless of the source database.

Backend support¶

Every language has at least one SQLite backend. Multi-engine backends (like java-jdbc, php-pdo, rust-sqlx) load engine-specific manifests automatically.

Type affinity system¶

SQLite uses type affinity rather than strict types. Any column can hold any type at runtime. Scythe maps declared types to neutral types based on the declared column type name.

CREATE TABLE users (
    id INTEGER PRIMARY KEY AUTOINCREMENT,
    name TEXT NOT NULL,
    score REAL,
    data BLOB
);

AUTOINCREMENT handling¶

CREATE TABLE users (
    id INTEGER PRIMARY KEY AUTOINCREMENT,
    name TEXT NOT NULL
);

INTEGER PRIMARY KEY is the SQLite auto-increment rowid. The AUTOINCREMENT keyword adds monotonicity. These columns are treated as NOT NULL int32.

Limitations¶

SQLite does not support:

• Enums -- no CREATE TYPE ... AS ENUM. Use TEXT with CHECK constraints instead.
• Arrays -- no array types. Use JSON arrays or separate tables.
• Schemas -- no schema.table syntax. Single namespace per database.
• Composite types -- no CREATE TYPE ... AS (...).
• Range types -- not available.
• Network types -- no INET, CIDR. Use TEXT.
• RETURNING -- only available in SQLite 3.35+ (2021).

Type mapping table¶

Placeholder syntax¶

SQLite uses $N positional placeholders, same as PostgreSQL:

SELECT id, name FROM users WHERE id = $1;