org.apache.drill.exec.physical.rowSet

Interface RowSet

All Known Subinterfaces:

RowSet.ExtendableRowSet, RowSet.HyperRowSet, RowSet.SingleRowSet

All Known Implementing Classes:

AbstractRowSet, AbstractSingleRowSet, DirectRowSet, HyperRowSetImpl, IndirectRowSet

public interface RowSet

A row set is a collection of rows stored as value vectors. Elsewhere in Drill we call this a "record batch", but that term has been overloaded to mean the runtime implementation of an operator.

A row set encapsulates a set of vectors and provides access to Drill's various "views" of vectors: VectorContainer, VectorAccessible, etc. The row set wraps a {#link TupleModel} which holds the vectors and column metadata. This form is optimized for easy use in testing; use other implementations for production code.

A row set is defined by a TupleMetadata. For testing purposes, a row set has a fixed schema; we don't allow changing the set of vectors dynamically.

The row set also provides a simple way to write and read records using the RowSetWriter and RowSetReader interfaces. As per Drill conventions, a row set can be written (once), read many times, and finally cleared.

Drill provides a large number of vector (data) types. Each requires a type-specific way to set data. The row set writer uses a ColumnWriter to set each value in a way unique to the specific data type. Similarly, the row set reader provides a ScalarReader interface. In both cases, columns can be accessed by index number (as defined in the schema) or by name.

A row set follows a schema. The schema starts as a BatchSchema, but is parsed and restructured into a variety of forms. In the original form, maps contain their value vectors. In the flattened form, all vectors for all maps (and the top-level tuple) are collected into a single structure. Since this structure is for testing, this somewhat-static structure works just file; we don't need the added complexity that comes from building the schema and data dynamically.

Putting this all together, the typical life-cycle flow is:

• Define the schema using SchemaBuilder.
• Create the row set from the schema.
• Populate the row set using a writer from RowSet.ExtendableRowSet.writer(int).
• Process the vector container using the code under test.
• Retrieve the results using a reader from reader().
• Dispose of vector memory with clear().

Nested Class Summary

Nested Classes

Modifier and Type	Interface and Description
static interface	RowSet.ExtendableRowSet Single row set which is empty and allows writing.
static interface	RowSet.HyperRowSet Row set comprised of multiple single row sets, along with an indirection vector (SV4).
static interface	RowSet.HyperRowSetBuilder
static interface	RowSet.SingleRowSet Row set that manages a single batch of rows.

Method Summary

Modifier and Type	Method and Description
BufferAllocator	allocator()
BatchSchema	batchSchema()
void	clear()
VectorContainer	container()
BatchSchema.SelectionVectorMode	indirectionType()
boolean	isExtendable()
boolean	isWritable()
void	print() Debug-only tool to visualize a row set for inspection.
RowSetReader	reader()
int	rowCount()
TupleMetadata	schema()
long	size() Return the size in memory of this record set, including indirection vectors, null vectors, offset vectors and the entire (used and unused) data vectors.
VectorAccessible	vectorAccessible()

Method Detail

isExtendable

boolean isExtendable()

isWritable

boolean isWritable()

vectorAccessible

VectorAccessible vectorAccessible()

container

VectorContainer container()

rowCount

int rowCount()

reader

RowSetReader reader()

clear

void clear()

schema

TupleMetadata schema()

allocator

BufferAllocator allocator()

indirectionType

BatchSchema.SelectionVectorMode indirectionType()

print

void print()

Debug-only tool to visualize a row set for inspection. Do not use this in production code.

size

long size()

Return the size in memory of this record set, including indirection vectors, null vectors, offset vectors and the entire (used and unused) data vectors.

Returns:

memory size in bytes

batchSchema

BatchSchema batchSchema()