[IDEA] Alternate model syntax

In trying to solve the DB → python type mapping, and the advent of typing, a possible solution is to define types that then get evaluated into both Python and DB types:

This also raises the question of a cleaner (more modern) model definition, by defining the type:

Since py3.6 we have the typed variables:

class Author(Model):
    name: str[255]
    books: reverse_fk[Book]

class Book(Model):
    title: str[255]
    author: Author
    published: datetime
    logo: bytes = None
    pages: int
    extra: JSONField = {}
    price: Decimal[18,5]
    created_at: datetime[auto_now_add]

Which is already standard in Py3.7 @dataclass

We could extend type definitions so we can build model definitions off of types.
e.g. str[100] = 'some default' → CharField(max_length=100, default='some default')
The syntax is not comprehensive, but we could probably use type declarations for a large portion of use cases.

This would require Py3.6+ to function, so dropping Py3.5 & PyPy.

Start by reviewing the proposed Python 3.6+ typing and dataclasses syntax alongside the existing model definitions. The issue names no files or tests, so the scope, supported declarations, database mappings, compatibility policy, and completion criteria would need to be established before implementation.