Skip to main content
Version: Next

Abstract Layer vs. Provider Projects

This page explains how the Abstract Backend library coordinates with provider repositories that deliver concrete queue implementations. Use it to visualise hand-offs between the shared abstraction, provider code, and consuming applications.

System view​

Responsibilities​

  • Abstract Backend (Core, Types, Loader, Consumer) – Defines stable protocols, discovery mechanisms, and runtime helpers. Ships in this repository.
  • Provider projects (Redis, SQS, Custom) – Independent packages that implement the protocols and register entry points. They own connection logic to their infrastructure.
  • Infrastructure (Redis Cluster, AWS SQS, etc.) – External services the providers interact with.
  • Consumers (your applications) – Import the abstract layer, select providers via environment variables, and use shared helpers to publish/consume events.

Collaboration workflow​

  1. Install – Applications install abstract-backend plus the desired provider package(s).
  2. Configure – Set QUEUE_BACKEND and provider-specific environment variables so load_backend() picks the correct implementation.
  3. Compose – Use abstractions like AsyncLoopConsumer to wire application handlers without referencing provider-specific APIs.
  4. Extend – Provider teams release updates independently; applications upgrade when ready without changing business logic.
  5. Monitor – Logging utilities (abe/logging/utils.py) give consistent visibility across providers.

This layered approach keeps application code stable while allowing backend capabilities to evolve via separate provider repositories.