Skip to content

redundant-optional

Default severity: ignore

What it does

Reports a field of a user-defined row type (or an item of a helper's tuple[...] row type) annotated T | None or Optional[T] while its query column cannot be NULL. This is the strict side of nullable-mismatch: that rule finds missing | None, this one finds extra.

[tool.sqlty]
schema = "schema.sql"

[tool.sqlty.rules]
redundant-optional = "warn"

schema.sql:

CREATE TABLE users (id INTEGER PRIMARY KEY, name TEXT NOT NULL);
from typing import TypedDict

from sqlty import SQL


class User(TypedDict):
    id: int
    name: str | None


# error: [redundant-optional] "User.name is str | None but query column 'name' is never NULL"
q = SQL[User]("--sql\nSELECT id, name FROM users")
Open in playground

A row type shared by several queries is judged by all of them: a field is reported only when no query that loads the type can return NULL for it. sqlty check gathers the queries of every checked file before it reports.

from typing import TypedDict

from sqlty import SQL


class Owned(TypedDict):
    id: int
    name: str | None


named = SQL[Owned]("--sql\nSELECT id, name FROM users")
maybe = SQL[Owned]("--sql\nSELECT u.id, o.name FROM users u LEFT JOIN users o ON o.id = u.id + 1")
# error:
Open in playground

Why is this bad?

Callers handle a None that never arrives, and the annotation hides which columns are really optional.

How to fix

Drop the | None, or accept it: a wider annotation is correct, just looser than needed.

from typing import TypedDict

from sqlty import SQL


class User(TypedDict):
    id: int
    name: str


q = SQL[User]("--sql\nSELECT id, name FROM users")
# error:
Open in playground

Configuration

Off by default, because it conflicts with a documented workflow: an expand and contract migration widens an annotation to T | None before the schema relaxes the column, and T | None on a non-null column is correct, only loose. Enable it:

[tool.sqlty.rules]
redundant-optional = "warn"

Only columns sqlty knows cannot be NULL are judged: a schema column with NOT NULL / PRIMARY KEY, a COUNT, a literal. A column of unknown nullability (a cast with no schema, an unknown function) is never reported. Any and object fields are not optional annotations and are not reported.