Release Notes¶

0.2.0¶

Live actor registry controls on ActorSystem, including refs(), refs_by_type(), broadcast(), and stop_all().
system.control_panel() for a printable snapshot of actor ids, addresses, names, actor classes, lifecycle state, and hierarchy.
ActorRef.is_alive() for lightweight liveness checks.
Async actor proxies via ref.proxy(), allowing method-style calls such as await proxy.reserve(...) while preserving actor mailbox isolation.
Fire-and-forget proxy method calls with await proxy.method.defer(...).
Direct proxy messages (ProxyCall, ProxyGetAttr, ProxySetAttr) for integrations that need lower-level control.
Real-world documentation for actor registries, control-panel usage, and actor proxies.

Persisted lifecycle event and dead-letter records now store stable actor address strings instead of leaking runtime-shaped address values.
Actor parent references now expose their address consistently through context.parent.address.
Watchers now receive one termination notification per stopped actor.
Startup persistence hooks are awaited correctly before actors are allowed to run.
Actor mailbox and internal event streams now close cleanly during shutdown and failed startup.
ActorAddress.parse() accepts the documented system://actor_id form while preserving the existing system:actor_id form.
Added the correctly spelled persistence_recovery property while keeping persistance_recovery as a compatibility alias.

When using the settings it was causing a conflict with the types and not casting properly to the right type due to the from future import __annotations__.
Settings inheritance now preserves typed fields from parent classes when type hint resolution falls back, allowing child settings classes to override inherited values without re-annotating fields.

Core async actor system with strict message isolation and single-threaded actor execution
ActorSystem lifecycle management (startup, shutdown, supervision)
Base Actor API with lifecycle hooks:
on_start
receive
on_stop
on_child_failure
Supervision model with restart, stop, escalate, and ignore strategies
Built-in persistence subsystem with pluggable backends:
In-memory backend
JSON file backend
Rotating file backend
Redis Streams backend (including consumer groups)
Retention policies based on:
Maximum record count
Maximum age
Maximum storage size
Compaction mechanisms to reclaim storage safely
Persistence health scanning and anomaly detection
Recovery strategies:
In-place repair
Quarantine-based recovery
First-class metrics system with:
Write/read counters
Error tracking
Recovery and compaction statistics
CLI tooling for operational control:
Persistence scan, recovery, compaction, and inspection
Doctor command for pre-flight health checks
Metrics inspection and reset
ASGI integrations with Lilya and Ravyn:
Automatic lifecycle hooks
Health endpoints
Metrics endpoints (JSON-compatible)
OpenTelemetry-compatible metrics exposure hooks

This is the initial public release of Papyra.
APIs are considered stable enough for early adopters, but minor breaking changes may occur before 1.0.0.
The project prioritizes correctness, observability, and operational safety over premature optimization.