Dataclass Interoperability

Spec-classes are interoperable with dataclasses, and data-class definitions can be easily upgraded into spec-class ones.

Here's a dataclass definition:

from dataclasses import dataclass, field

@dataclass
class Farm:
    owner: str = "Harry"
    animals: List[str] = field(default_factory=list)

f = Farm()
f # Farm(owner='Harry', animals=[])

# We're not protected here from any abuse of types at run-time.
f.owner = 10
f # Farm(owner=10, animals=[])

Let's upgrade that to a spec-class the easiest possible way:

from spec_classes import spec_class
from dataclasses import field

@spec_class
class Farm:
    owner: str = "Harry"
    animals: List[str] = field(default_factory=list)

f = Farm()
f # Farm(owner='Harry', animals=[])

# Looks the same! But now we're enforcing types.

f.owner = 10
# TypeError: Attempt to set `Farm.owner` with an invalid type [got `10`; expecting `str`].

Spec-classes also has its own equivalent of field that provides some additional spec-classes specific configuration (see Attr), so the above is equivalent to:

from spec_classes import spec_class, Attr

@spec_class
class Farm:
    owner: str = "Harry"
    animals: List[str] = Attr(default_factory=list)

Spec-classes is also special in that it allows mutable types to be set as the defaults on base class definition (they are safely copied at instantiation time in the generated constructor). Thus, you can actually just write:

from spec_classes import spec_class

@spec_class
class Farm:
    owner: str = "Harry"
    animals: List[str] = []

Boom! Done.