Reference¶
Reference material is for exact behavior after the shape of a weave is already clear: public imports, configuration keys, environment resolution, executor frames, and backend adapter requirements. The topic guides explain the workflow model; the reference pages pin down names, options, and runtime contracts.
Core References¶
Public API: the import surface most users should rely on.
Environments: how persistent execution profiles are defined and resolved.
Executors: executor contracts, error frames, dispatch extras, network executors, and custom executor extension points.
Backend Adapters: setup for Wove’s built-in bridges into external task systems and custom backend adapter extension points.
Public API¶
The public API references separate the stable import surface from modules that mainly matter when inspecting internals.
wove: package entrypoints includingweave,Weave,config, helpers, and exported exceptions.wove.weave: reusable workflow classes and inheritable weave definitions.wove.helpers: data-shaping helpers used inside task graphs.wove.context: context manager implementation behindwith weave() as w:.
Configuration And Runtime¶
Runtime references trace how Wove resolves project defaults, named environments, task-level overrides, and executor instances when work is not running where expected.
wove.runtime: process-wide configuration singleton andwove.config(...)behavior.Environments: environment dictionaries, precedence, defaults, and validation rules.
wove.environment: executor interfaces, runtime delivery errors, and executor runtime classes.wove.security: network executor request signing and verification helpers.
Dispatch And Remote Execution¶
Remote execution references separate the direct executor protocol, worker-service flow, backend callback flow, and adapter contract so the two extension shapes do not blur together.
Executors: frame protocol, executor lifecycle, and network executors.
HTTP/HTTPS Executor: direct request/response transport for worker services.
gRPC Executor: generic unary RPC transport for gRPC-based worker services.
WebSocket Executor: persistent bidirectional transport for worker services that stream events.
Custom Executors: direct executor implementations for project-owned transports or runtimes.
Backend Adapters: built-in bridges into Celery, Temporal, Ray, Kubernetes, AWS Batch, Slurm, and similar systems.
Custom Backend Adapters: adapter implementations for project-owned task systems.
wove.backend: backend callback transport and dispatch payload helpers.wove.integrations: adapter registry, adapter base interface, and worker entrypoints.wove.security: shared security layer for HTTP, gRPC, and WebSocket network executors.
Backend Adapters¶
Each adapter page shows what Wove submits to that system, what the worker must run, and which executor_config keys matter.
Celery: broker-backed worker pools.
Temporal: workflow/task-queue execution.
Ray: Ray cluster tasks.
RQ: Redis Queue workers.
Taskiq: async task queues with explicit task registration.
ARQ: async Redis worker functions.
Dask: distributed Python scheduler workers.
Kubernetes Jobs: isolated pod-per-task execution.
AWS Batch: managed batch compute jobs.
Slurm: HPC batch scheduling.
Custom Backend Adapters: project-local bridges into systems Wove does not ship with.