软件包 ml.shifu.guagua.io

This package contains our own io settings for guagua.

请参见：
描述

接口摘要

Bytable	Copy from hadoop io writable.
GuaguaRecordReader<KEY extends Bytable,VALUE extends Bytable>	GuaguaRecordReader is used for consistent interface to iterate data through FileSplit provided.
Serializer<RESULT extends Bytable>	Serializer is used to serialize object to bytes and de-serialize bytes to object A CONTRACT here: To make sure halt status in Bytable can be serialized or de-serialized.

类摘要

BytableSerializer<RESULT extends Bytable>	BytableSerializer is usingBytable.write(java.io.DataOutput) and Bytable.readFields(java.io.DataInput) to serialize object to bytes and de-serialize bytes to object.
Bzip2BytableSerializer<RESULT extends Bytable>	Bzip2BytableSerializer is usingBytable.write(java.io.DataOutput) and Bytable.readFields(java.io.DataInput) to serialize object to bytes and de-serialize bytes to object.
GuaguaFileSplit	GuaguaFileSplit is used for WorkerComputable to read data.
GZIPBytableSerializer<RESULT extends Bytable>	GZIPBytableSerializer is usingBytable.write(java.io.DataOutput) and Bytable.readFields(java.io.DataInput) to serialize object to bytes and de-serialize bytes to object.
HaltBytable	Abstract Bytable implementation to read and store halt status.

软件包 ml.shifu.guagua.io 的描述

This package contains our own io settings for guagua.

com.paypal.guagua.io.Bytable is a copy from hadoop Writable, the reason is that we don't want to depend on hadoop platform since guagua is designed to support all kinds of computation platform. One can check GuaguaWritableAdapter in guagua-mapreduce to see how to use hadoop writable with guagua com.paypal.guagua.io.Bytable.

Compressed serialzer like com.paypal.guagua.io.GZIPBytableSerializer is used to support gz format. And com.paypal.guagua.io.Bzip2BytableSerializer is for bz2 format.