We’re gradually working towards a 3.10 release. The headline feature here will be proper OpenAPI schema generation. We’re pulling out the intermediary CoreAPI document-model approach that we’ve been using until now, since OpenAPI has now long become a widely adopted standard for modelling Web APIs.
Alongside this there’s been work on the stand-alone apistar tooling, which provides API documentation generation, schema validation, and a schema-driven client library.
We’re continuing to push forward the async landscape, since it represents a huge potential for Python.
The requests-async package brings async/await support to the ever-popular requests library.
There are currently a couple of constraints on the package - most notably we don’t yet have streaming upload/download support. We’ll likely invest some time in the coming month into resolving any outstanding feature-differences between requests and requests-async.
The new orm package is an async ORM with a Django-like API.
Because it is built on top of databases it has support for SQLite, Postgres, and MySQL. Query building is based on SQLAlchemy core, which means we’re also able to provide migration support, through Alembic.
The ORM package is still in the development stage, but has a fairly fully-featured API.
The stack of async functionality that we now have expands all the way through from an ASGI server implementation, all the way up to an ORM.
uvicorn - Web Serverstarlette - Web frameworkdatabases - Low level cross-database queriesorm - High level cross-database ORMrequests-async - ASync HTTP requestsEach of the pieces in the stack are necessary pre-requisites to Python offering a mature, fully featured, high-performance async web stack.
I’ll be talking about the work here, the payoff it’ll enable, and where Django might fit in with this, at DjangoCon Europe, next week.
As ever thank you so much to all our sponsors, contributors, and users.
— Tom Christie, 3rd April, 2019.