FluentKit 1.0.0 GM

[tanner0101]

Given the large number of fixes currently on the gm branch, I've decided to release them as a final rc version now. The official 1.0.0 will release will come as soon as the docs for Fluent are finished.

---

Unify field key path semantics (#268).

Makes FieldKey path usage consistent across all db driver types. SQL no longer treats field key paths differently than NoSQL. This means @Group now always serializes as a flat structure. Support for nested structures and dynamic structuring behavior between NoSQL and SQL will be added in the future with a more reliable design.

Improve property protocols (#268).

FieldProtocol and PropertyProtocol have been cleaned up and split into more descriptive protocols:

• AnyProperty
• AnyDatabaseProperty
• AnyCodableProperty
• AnyQueryableProperty
• Property
• QueryableProperty

Add Database.inTransaction (fixes #23, #119).

Database drivers are expected to keep track of whether they are already in a transaction and not start a new one if they do not supported nested transactions. Users can check if the database they are using is currently in a transaction using the inTransaction flag.

Fix issues filtering and setting @OptionalField when nil (#270).

@OptionalField.value is now a double optional again. This was previously changed to avoid double optionals which can be tricky to work with, but a single optional does not accurately convey all states an @OptionalField can be in:

• nil: Uninitialized
• .some(nil): Initialized to nil.
• .some(.some): Initialized to a value.

Furthermore, @OptionalField's associated value type being an optional is required for generic operators (like ==) and methods (like set(_:to:) to properly accept nil values.

Add batch delete method on collection of models (#270, fixes #114).

A new method delete(force:on:) is now available on collections of models.

let earth: Planet = ...
let mars: Planet = ...
[earth, mars].delete(on: db)

Schema update constraints (#270, fixes #235).

Schema constraints can now be created and deleted during SchemaBuilder.update().

database.schema("planets")
    .unique(on: "name")
    .update()

Add constraint names (#270, fixes #118).

New methods on SchemaBuilder are now available for naming unique and foreign key constraints.

database.schema("planets")
    .field("name", .string, .required)
    .unique(on: "name", name: "name_index")

Constraints can now be deleted by name as well.

database.schema("planets")
    .deleteConstraint(name: "name_index")

Arbitrary join(_:on:) ordering (#270, fixes #128).

QueryBuilder.join's on parameter can now have fields appear on either side of ==.

School.query(on: self.database)
    .join(City.self, on: \School.$city.$id == \City.$id)
    // or 
    .join(City.self, on: \City.$id == \School.$city.$id)

Add EnumBuilder.read() (#270, fixes #194).

EnumBuilder now supports reading the enum metadata outside of create(). This is useful for re-using enums between schemas.

database.enum("planet_type").read().flatMap { planetType in
    database.schema("planets")
        .field("type", planetType)
        .create()
}

Improve readability of Codable errors (#270, fixes #231).

DecodingErrors thrown by Model now include key path information.

Several small bug and convenience API fixes (#270).

[tanner0101]

These changes are now available in 1.0.0-rc.2