Zoo Framework

Python multithreading framework

Long-lived background tasks in your own process — scheduled, observable, stateful, with no broker and no cron daemon

Worker — task execution unit

Each task is an animal — loop-based, period-based, event-driven or state-machine driven, each on its own beat

Process / session / prototype scopes hold shared instances — declarative, resettable, replaceable, and never replacing the class itself

Loads config, registers workers, drives the dispatch loop, and shuts down gracefully and in order on Ctrl-C

Events flow through channel registration, FIFO queues and reactors, with priorities and response mechanisms to avoid starvation

StateScopes sit on pluggable StateIndexes and are periodically saved to a pickle archive with checksum and rolling backups

Plugin ABC with dependency-ordered loading, plus fixed / exponential / adaptive delay strategies