v0.4.0

• Context factory: Context factor for injection of custom-made context classes

v0.3.0

• Fixed critical caching bug: manifest.json is now only written on successful task completion, ensuring that failed tasks can be safely retried.
• Dedicated error state: Added a new error.json marker file written upon task failure, capturing the error message, traceback, and a partial_manifest of files written prior to failure.

v0.2.0

Breaking changes

• Renamed "workflow" to "task" across the entire codebase (Workflow -> Task, @register_workflow -> @register_task, etc.).
• Removed TaskContextProtocol. Use TaskContext directly instead.
• Caching moved to execution time: Pipeline.build_plan() no longer accepts an is_cached callback. Cache detection now happens automatically inside execute_task() — if manifest.json already exists at the task's storage prefix, the task is skipped without re-running. Remove any is_cached=... arguments from build_plan() calls and any LocalStorageBackend.make_cache_checker() / is_result_cached() usage.
• TaskNode.cached field removed. Tasks are no longer pre-marked cached at plan-build time.
• run_plan_locally() no longer accepts a use_cache parameter. Caching is always active and requires no configuration.
• LocalStorageBackend.make_cache_checker() and LocalStorageBackend.is_result_cached() removed.
• ExecutionResult gains a new cached: bool field (default False) indicating whether the task was skipped due to an existing result.
• ExecutionResult.files_written removed. Tasks communicate results exclusively through file writes to the storage context.
• submit_plan() now returns PlanHandle instead of a raw str. Update any code that stores or inspects the return value directly.
• on_node_start, on_node_complete, and on_node_failure removed from the ExecutionBackend protocol. They remain as keyword-only parameters on LocalBackend.submit_plan() for testing and CLI use.
• CompletionCallback.notify() signature changed from (analysis_id: int, result: ExecutionResult) to (plan_id: str, success: bool, error: Optional[str]). Update any custom callback implementations.

New features

• save_text / read_text: StorageBackend, LocalStorageBackend, S3StorageBackend, and TaskContext now expose save_text(filename, data, encoding="utf-8") for writing plain-text files, complementing the existing save_file (bytes), save_json, and save_xarray. TaskContext also adds read_text(filename, encoding="utf-8") as a convenience wrapper. This closes the gap with OutputFile(file_type="text"), which had no corresponding write method.
• open_file write-mode guard: open_file is read-only by design. Both LocalStorageBackend and S3StorageBackend now raise ValueError if a write mode (w, a, or x) is passed. Previously LocalStorageBackend silently opened the file for writing while bypassing write-once semantics, allowed_outputs checks, and manifest tracking; S3StorageBackend silently discarded the write and returned a read buffer instead.
• PlanHandle: serializable reference to a submitted plan execution. Call handle.to_json() to store it (e.g. in a Django model field) and PlanHandle.from_json(s) to restore it later. Methods:
  • get_state() -> str — "pending" | "running" | "success" | "failure". For Celery hits Redis; for Step Functions calls describe_execution; never queries S3.
  • get_progress() -> PlanProgress — per-node completion based on manifest.json presence.
  • cancel() — revokes the Celery chord or stops the Step Functions execution.
  • configure_celery(app) — class-level method; call once at startup so get_state() / cancel() work outside the CeleryBackend instance.
• PlanProgress: returned by PlanHandle.get_progress(). Fields: total, completed, node_breakdown: dict[str, bool]. Properties: fraction (0.0–1.0), is_complete.
• ProgressChecker protocol (muflow.storage): storage-layer abstraction for checking node completion across multiple prefixes without querying an execution backend.
  • LocalProgressChecker: checks os.path.exists(prefix/manifest.json).
  • S3ProgressChecker(bucket): one HEAD request per prefix (10–50 ms each within the same AWS region).
  • make_progress_checker(storage_type, storage_config): factory that reconstructs the right implementation from serialized config; extend by adding a new class and one branch here.
• CeleryCompletionCallback wired into CeleryBackend: pass completion_callback=CeleryCompletionCallback(app, "myapp.tasks.on_complete") to submit_plan() and a muflow.send_completion Celery task fires after the plan finishes, calling your task with (plan_id, success, error). The muflow.send_completion task is registered automatically by create_celery_task().

v0.1.0

Initial release.

Features

• Workflow registry: @register_workflow decorator for registering pure
  workflow functions with optional Pydantic parameter validation
• Pipeline abstraction: Pipeline, Step, and ForEach for declarative
  multi-step DAG definitions
• WorkflowPlan: Static, serializable DAG representation compiled from
  pipelines via topological sort
• Content-addressed storage: Deterministic prefix computation with
  IdentityKey annotations for cache control
• WorkflowContext: Unified file I/O interface (JSON, xarray, raw bytes)
  with dependency access and progress reporting

Execution backends

• LocalBackend: Synchronous in-process execution for testing and CLI use
• CeleryBackend: Parallel DAG execution via Celery chord/group primitives
• StepFunctionsBackend: AWS Step Functions orchestration with Lambda

Storage backends

• LocalStorageBackend: Filesystem-based storage with write-once semantics
  and path traversal protection
• S3StorageBackend: AWS S3 storage backend

Utilities

• run_plan_locally() helper for pipeline integration testing
• ResourceManager and resolve_uri for transparent local/remote resource
  fetching
• Extended JSON encoder with NaN, numpy, and datetime support
• xarray Dataset serialization helpers