org.apache.drill.exec.physical.resultSet

Interface PushResultSetReader

All Known Implementing Classes:

PushResultSetReaderImpl

public interface PushResultSetReader

Push-based result set reader, in which the caller obtains batches and registers them with the implementation. The client thus is responsible for detecting the end of batches and releasing memory. General protocol:

• Create an instance and bind it to a batch source.
• Obtain a batch, typically by having it passed in.
• Call start() to obtain a reader for that batch.
• Iterate over the rows.
• Release memory for the batch.

In Drill, batches may have the same or different schemas. Each call to start() prepares a RowSetReader to use for the available batch. If the batch has the same schema as the previous, then the existing reader is simply repositioned at the start of the batch. If the schema changed (or this is the first batch), then a new reader is created. Thus, the client should not assume that the same reader is available across calls. However, if it is useful to cache column writers, simply check if the reader returned from start() is the same as the previous one. If so, the column writers are also the same.

Method Summary

Modifier and Type	Method and Description
RowSetReader	start()

Method Detail

start

RowSetReader start()