The main idea being tested is how threading works in a flow that processes requests in a blocking, single-threaded manner. In a synchronous flow, the entire processing, including creating the response, happens on the same thread from the moment the request is received until the response is sent back. It uses only the message source thread pool, with no offloading to other threads. This is why the best description is that all processing, including the response, is done in the same thread and relies solely on the inbound (message source) thread pool. It ensures a straightforward, blocking execution path from start to finish. The other scenarios describe asynchronous or offloaded work—processing on multiple threads, only part of the flow using a thread pool, or using queues by default—which do not define a synchronous flow.

During a batch job, the Load phase gathers all the records, and the Dispatch phase places those records into a queue so multiple workers can pull items off and process them in parallel. A queue fits this role because it provides a simple, orderly way to distribute work in FIFO order as resources become available. This is ideal for processing many records efficiently across parallel steps. Other structures don’t match this use: a Map is for key-value pairs, a Set enforces unique elements without a processing order, and an Array is just a static list without built-in mechanisms to feed multiple workers.

The watermark from the Poll scope is exposed as a flow variable, so it can be read by any processor inside the flow using the flowVars map. Inside a processor you’d typically access it with an expression like #[flowVars.watermark], which retrieves the last poll position. This keeps the poll's state local to the flow instance and avoids mixing with message headers or flow properties, which serve different purposes. If you see a different variable name for the watermark in your flow, use that name, but the mechanism remains: read the value from a flow variable rather than a property or header.

DataWeave 1.0 XML output requires a single root element, defined by the top-level key followed by a colon and a map of its children. The structure rootName: { ... } tells DataWeave to create an XML document with a root element named rootName, whose child elements come from the keys inside the nested map. In this example, using user as the root yields a single element containing John and Doe, which is a valid XML with one root. The other forms would not produce a single root element. If you had multiple top-level keys, you’d end up with multiple root elements, which XML does not allow. A literal XML string shown is not the DW root expression that generates XML from a data structure.