Python dataclasses are syntactic sugar over ordinary classes: the `@dataclass` decorator generates boilerplate (initializer, repr, comparison methods) from declared fields, but the resulting object retains full normal-class semantics, so it can be composed with class variables, class methods, decorators, descriptors, `__post_init__`, dunder methods (`__enter__`/`__exit__`), and runtime field introspection via the `dataclasses.fields()` API. This composability enables design patterns — singleton/memoized construction via cached class-variable state, self-registration via decorator composition, runtime validation frameworks, schema derivation from field metadata, lazy cached computed attributes, declarative CLI-argument parsing, and context-manager resource wrappers — that belong to the broader discipline of Python object-oriented design and metaprogramming rather than to dataclasses as a standalone feature. A key formal mechanism is `InitVar`, which declares a value accepted by the generated initializer and passed to `__post_init__` without being retained as an instance field, distinct from ordinary fields which are both initializer parameters and persisted attributes.
Python dataclasses are syntactic sugar over ordinary classes: the `@dataclass` decorator generates boilerplate (initializer, repr, comparison methods) from declared fields, but the resulting object r…