Type safety

Type safety is the property of a type system that prevents, at parse-time or compile-time, the use of values incompatible with the declared type. In a database context, it means the engine rejects operations that violate type constraints before even executing the query.

How it works #

When a column, function parameter or variable is declared with a specific type (e.g. INTEGER, a custom domain, an ENUM), the engine checks at parse time that every value assigned or compared is compatible. Operations that mix incompatible types without an explicit cast raise an error before execution, preventing bugs that would otherwise only show up at runtime.

What it’s for #

To shift error detection from runtime to parse-time, reducing the risk of corrupt or inconsistent data in production. In PostgreSQL, for instance, an ENUM is a standalone type: a function that takes a subscription_status cannot be called with an arbitrary string. In MySQL, where ENUM is just a decoration of a VARCHAR column, this guarantee doesn’t exist — the constraint is on the column, not on the type itself.

When to use it #

It’s useful in any system where semantic data integrity matters more than ease of writing: billing, finance, customer records, any domain where an “out-of-domain” value is a business error rather than an acceptable variant. End-to-end type safety is one of PostgreSQL’s hallmark traits and a major reason it gets picked for enterprise workloads.