Composable pipeline architecture built from Python generators, where each stage is a function taking an iterable and returning a generator, enabling lazy evaluation and consumer-driven pull semantics rather than push-based data flow. The core mechanism is backpressure: because each stage only produces a value when the downstream consumer requests one (via `next`), execution demand propagates upstream through the whole chain, bounding memory use and avoiding intermediate buffering. This extends to bidirectional communication via `send` (stateful generators that receive values), generator return values (captured via `StopIteration`), and async generators, which apply the same pull-based, lazy-evaluation model to asynchronous/awaited data sources.
Composable pipeline architecture built from Python generators, where each stage is a function taking an iterable and returning a generator, enabling lazy evaluation and consumer-driven pull semantics…