Changelog — CentauroDB

v0.9.0

First Public Release

CentauroDB lands on PyPI: Pydantic-native, zero-migration persistence on SQLite and PostgreSQL, published under MIT for Python 3.11+. The SQLite path is ready for real workloads; PostgreSQL runs the same API on a single connection while pooling is in the works.

Published on PyPI: pip install centaurodb (Python 3.11+, MIT)

count_objects() and aggregate(count=True) for totals without materializing rows

limit / offset pagination on read_objects()

parent_ref(on_delete="restrict"): deleting a parent with children now raises ParentHasChildrenError instead of orphaning rows

register_backend() and engine.async_safety for pluggable backends

PostgreSQL integration tests in CI; fully type-hinted (py.typed)

Pre-release internal history — never published to PyPI

v0.5.0

Raw SQL Escape Hatches & Deletion API

Full CRUD coverage with delete methods, raw SQL queries via sql_select() and sql_execute(), read_latest_object() with conditions, and async parity across the board.

read_latest_object() with conditions and aggregates

sql_select() and sql_execute() for raw SQL escape hatches

delete_object() / delete_objects() methods

Improved index workflow

Async collection methods at parity with sync API

v0.4.0

Query API & Relationships

Powerful AND/OR query conditions on JSON fields, selective field projection in views, index support on views and tables, and 1-to-many relationships between collections.

Query API with AND/OR conditions on JSON fields

Selective field projection in views

Index support on both views and tables

1-to-many relationships between collections

v0.3.0

Async & PostgreSQL Support

AsyncCollection and AsyncTimeSeriesCollection wrappers, PostgreSQL backend via Engine URL dispatch, dialect system for backend-agnostic SQL, and batch DataFrame validation.

AsyncCollection and AsyncTimeSeriesCollection wrappers

PostgreSQL backend via Engine URL dispatch

Dialect system: SQLiteDialect / PostgresDialect

Batch DataFrame validation for time-series

v0.2.0

Views, Time Series & Zero-Migration DX

SQL views with JSON field extraction, full TimeSeriesCollection with Polars DataFrames, refresh() workflow, and backward-compatible field renames via renamed_from().

create_view() with JSON field extraction and expression indexes

Full TimeSeriesCollection with Polars DataFrame support

refresh() workflow to re-read objects from DB

renamed_from() for safe field renames

Backward compatibility guardrails

v0.1.0

Initial Release

Engine and Collection with SQLite backend, CentauroModel based on Pydantic v2, core CRUD workflow, BaseCollection/TimeSeriesCollection split, and custom exception hierarchy.

Engine and Collection with SQLite backend

CentauroModel based on Pydantic v2

write_object(), read_object(), update_object() CRUD

Custom exceptions for persistence tracking