Django Nova

A typed, unified, async-first toolkit for Django 5+.

Django Nova site

Django Nova Documentation

Welcome to the documentation of Django Nova— a typed, unified, asynchronous-oriented Django toolkit. The library is designed with a focus on scientific computing, Highload, and Reproducible Research.

What is Django Nova?

Django Nova is a modern toolkit that solves Django's key architectural problems.:

• Duplicate validation — you no longer need to write validation in forms, serializers, and models separately
• No strict typing — full support for pyright --strict
• Caching Issues — Smart disability without manual control
• Difficulties with migrations — built-in support for PostgreSQL concurrent migrations

Project philosophy

1. A single source of truth — all business logic of validation is concentrated in Pydantic schemes
2. Fail fast — errors should be detected at the static analysis stage, not in runtime
3. Default asynchrony — all operations are designed with asyncio in mind
4. Zero-downtime — migrations and updates should not interrupt the system operation

Modules

nova.typing

A strict typing layer. Includes NovaModel' and NovaConfig'. Uses PEP 695 to ensure full type derivability in the IDE (PyCharm, VSCode + Pyright).

nova.validation

Django Unified Bridge <-> Pydantic (pydantic_bridge). Ensures that validation rules are not duplicated between forms, serializers, and models.

nova.cache

Intelligent QuerySet caching ('queryset_cache'). Features:

• Using SQL Compiler to generate hashes (safe with Django updates).
• Reversible index O(1) for instant cache invalidation during save() or `delete()'.

nova.tasks

Built-in asynchronous task engine based on asyncio.Queue`. An alternative to Celery for in-process computing (ML inference, simulation).

nova.db

Utilities for secure migrations:

• zero_downtime.py : Wrappers over CREATE INDEX CONCURRENTLY and `ALTER TABLE' without locks (PostgreSQL).
• splitter.py : Breaking down heavy Data Migrations into batches to prevent OOM.

📚 More Information

• Full Documentation: docs/index.md
• Auto-generated Status Report: STATUS.md
• Changelog & Release Notes: CHANGELOG.md

🚀 Quick Start (5 minutes)

1. Create a typed model

## src/app/models.pyfrom django.db import modelsfrom nova.typing import NovaModel, NovaConfigfrom pydantic import BaseModelclass ArticleSchema(BaseModel):    title: str    content: str    views: int = 0class Article(NovaModel):    _nova_config = NovaConfig(        pydantic_schema=ArticleSchema,        strict_validation=True,        cache_enabled=True,    )    title: models.CharField(max_length=200)    content: models.TextField()

2. Use it with automatic validation

# Views or services
article = Article(title="Hello Nova", content="Typed Django!")
article.save()  # ✅ Validates against Pydantic schema automatically

3. Enjoy type safety 🎉

# Pyright --strict compatible! ✅
article.title = 123  # Type error! Expected str

📦 Installation

Requires Python 3.12+ and Django 5.0+ (tested with Django 5.0, 5.1, 5.2).

Using uv (recommended):

# Core library
uv add django-nova

# With Django REST Framework support
uv add django-nova[drf]

# With Redis infrastructure & Distributed Locks
uv add django-nova[redis]
# or
uv add django-nova[cache]

# With OpenTelemetry tracing
uv add django-nova[tracing]

# Full enterprise stack (tracing + observability)
uv add django-nova[tracing,observability]

# With FastAPI integration
uv add django-nova[fastapi]

# With async task queue
uv add django-nova[tasks]

# With async database support
uv add django-nova[async]

Add to INSTALLED_APPS:

INSTALLED_APPS = [
    # ...
    "nova",
]

⚡ Performance

Nova adds zero overhead compared to pure Django:

Key insights:

• ✅ Nova is faster than pure Django (within measurement noise)
• ✅ Django adds 2.65x overhead vs Pydantic (ORM vs dataclass)
• ✅ Nova adds zero additional overhead on top of Django

Note: Nova → Pydantic conversion costs 7.824 µs/iter, but this is only incurred when serializing to API responses or performing cross-layer validation.

Benchmark Details

uv run python scripts/bench.py

Runs 100,000 iterations with GC disabled to measure real algorithmic performance.

🎯 Philosophy

⚠️ This is a Beta project. See the auto-generated status report for real module-by-module coverage and stability assessment.

📊 Honest Project Status

🚀 Quick Start

📊 Current Status

🛡️ Validation Boundary

See STATUS.md for the honest, auto-generated breakdown.

📚 API Reference

👤 Author

Artem Alimpiev

• ORCID: 0009-0007-6740-7242
• DOI: 10.5281/zenodo.20057443
• DOI: 10.5281/zenodo.20659647
• PyPI: Django Nova
• NOVA: Django Nova site

📄 License

MIT License. See LICENSE for details. 2026