Runtime - BlockX Docs

The runtime carries your function computation from submission to result. Workers run tasks, Python Executors execute function code, and backend adapters provide table and EVM node access.

Choose an entry point

Entry	Use it for	Result handling
Worker task	Several function calls from explicit inputs, a block, or a bundle	Watch the task and inspect its Writer Plugin result
Sync Invoker	One function call or a debug session	Read the response of `Invoke` or `DebugInvoke`
Precheck	Check function source before execution	Read audit findings without executing the function

Choose the deployment

Deployment	Workload
block	Per-block computation and returned task outputs
bundle	Historical bundle computation and batch writes

Choose a deployment that registers the required Call Builder and Writer Plugin. See configuration reference for their wire shapes.

Prepare dependencies

Explicit pure-computation inputs need no chain data.
BlockDB reads need accessible source tables and the required data readiness.
EVM state reads and contract calls need a node backend serving the requested chain and block.
Historical scans need the source bundle data; writes need a compatible target table.

Set the read context in the function rather than relying on a deployment to supply it. A local smoke test proves the execution path; connect the required backends before using EVM examples.

Task connection

A Coordinator can select a Worker and reserve a slot. Submit to that Worker and keep WatchTasks connected while the task runs. GetTaskResult queries the same Worker’s stored result.

Summary

Choose task execution or a synchronous function call.
Match the deployment to the input and output configuration.
Provide the EVM and table backends required by the computation.

Continue reading: