org.apache.spark.sql.execution.columnar.encoding
BooleanRunLengthDecoder
Related Doc:
package encoding
org.apache.spark.sql.execution.columnar.encoding
any different handling when encoded positions end (e.g.
any different handling when encoded positions end (e.g. for nulls, after last null)
The next position at which the value changes.
The next position at which the value changes. A value < Int.Max / 2 indicates that current run is of trues while a value greater than that indicates that it is of false values (the maximum number of values in a column batch cannot exceed Int.Max/2).
Reads the boolean value at given position.
Reads the boolean value at given position. Calls to this method should be monotonically increasing and never decrease else result may be incorrect.
Read short run-length encoded as 1 or 2 bytes.
Read short run-length encoded as 1 or 2 bytes.
Run length encoding optimized for booleans. Each short run-length value represents the run-length with the value. Even numbered run-lengths indicate a run of false values having half the length, while odd numbered run-lengths are for true values (having length = run / 2 + 1).