v0.1.9

v0.1.9 (2026-01-05)

New Features

Management API for Instance Deletion and Pruning

This release introduces comprehensive instance lifecycle management APIs:

Client API

• delete_instance(id, force) - Delete single instance with cascading
• delete_instance_bulk(filter) - Bulk delete with filters (IDs, timestamp, limit)
• prune_executions(id, options) - Prune old executions from long-running instances
• prune_executions_bulk(filter, options) - Bulk prune across multiple instances
• get_instance_tree(id) - Inspect instance hierarchy before deletion

Provider API (ProviderAdmin trait)

• delete_instance(id, force) - Provider-level single deletion
• delete_instance_bulk(filter) - Provider-level bulk deletion
• delete_instances_atomic(ids) - Atomic batch deletion for cascading
• prune_executions(id, options) - Provider-level pruning
• prune_executions_bulk(filter, options) - Provider-level bulk pruning
• get_instance_tree(id) - Provider-level tree traversal
• list_children(id) - List direct child sub-orchestrations
• get_parent_id(id) - Get parent instance ID

Safety Guarantees

• Running instances protected (skip or error based on API)
• Current execution never pruned
• Sub-orchestrations cannot be deleted directly (must delete root)
• Atomic cascading deletes (all-or-nothing)
• Force delete available for stuck instances

Testing

• 102 new provider validation tests (deletion, bulk deletion, pruning)
• Integration tests for all Client methods
• Cascading delete tests (parent→children)
• Filter combination tests (IDs, timestamps, limits)
• Safety tests (running protection, current execution protection)

Documentation

• Provider implementation guide with contracts
• Provider testing guide updates
• Continue-as-new docs with pruning section
• README instance management section
• Enhanced management-api-deletion proposal with force delete semantics

Other

• Updated release checklist prompt
• README latest release link updated

v0.1.1: Long Polling Support

What's Changed

This release adds long polling support to the Provider interface and makes continue_as_new() awaitable.

Breaking Changes

• Provider trait: Added poll_timeout: Duration parameter to fetch_orchestration_item and fetch_work_item
• RuntimeOptions: Renamed dispatcher_idle_sleep to dispatcher_min_poll_interval, added dispatcher_long_poll_timeout
• continue_as_new(): Now returns an awaitable future (use: return ctx.continue_as_new(input).await)

New Features

• Long polling support for providers (Redis, Azure Service Bus, etc.)
• dispatcher_long_poll_timeout configuration (default: 5 minutes)
• Long polling validation tests in provider_validations module

Migration Guide

See CHANGELOG.md for detailed migration instructions.

Full Changelog: v0.1.0...v0.1.1