Architecture overview

BlockX connects EVM data semantics to a task execution system. Call Builders select input, Python Executors run user functions, Worker IO routes dependency reads, and Writer Plugins submit derived outputs.

​

Responsibilities and data path

The caller supplies the intended input and output context. A function explicitly selects the chain and block for its EVM reads. BlockDB handles data storage; node capabilities execute contract queries; BlockX schedules and runs the surrounding function computation.

​

Execution flow

1. The caller reserves capacity through a Coordinator or submits directly to a Worker.
2. The Worker accepts the task and activates its execution context.
3. The Call Builder reads or expands the configured input into function calls.
4. Python Executor runs each function; backend operations pass through Worker IO.
5. The Writer Plugin handles collected outputs and the Worker records the terminal result.

For large bundle input, call production and execution can overlap. The output phase still follows the collected computation. Task lifecycle explains the state transitions.

​

Component map

​

Completion boundary

The Worker reports the task’s result-handling outcome. A block write submits asynchronously; a bundle write returns after BatchWrite commit submission. Downstream consumers check storage readiness separately. The common contract lives in execution semantics.

​

Contributor reading order

• Find the relevant module in Repository layout.
• Read Design principles and the component’s boundaries.
• Follow Testing and Contributing.

For product vocabulary, start with EVM computation model. For request fields, use Service API.