CODEWITHSUNDEEP
javahadoopmapreduce
diplomaticguru
diplomaticguru

Reputation: 675

Why do we need to explicitly set OutputKey/InputKey Class in MapReduce job?

Why do we need to explicitly set OutputKey/InputKey Class in MapReduce job when the Map and Reduce classes has to extend Mapper and Reducer classes respectfully that requires to add <InputKeyClass, InputValueClass, OuputKeyClass, OutputValueClass> ?

an example:

JobConf:

    job.setOutputKeyClass(NullWritable.class);
    job.setOutputValueClass(Text.class);
    job.setOutputFormatClass(TextOutputFormat.class);

Reducer:

public static class ComputeReducer extends
        Reducer<Text, Text, NullWritable, Text> {

If we are already explicitly setting the reducer class to extend 'Reducer <Text, Text, NullWritable, Text>' then why is it required to pass the OutputKey/value class?

Upvotes: 0

Views: 71

Answers (1)

Todoy
Todoy

Reputation: 1266

You have to do that because all generic information is lost at run time in java.

Upvotes: 3

Related Questions