MapWritable works if you use a binary format like
SequenceFileOutputFormat. If you need a plain text output, you can use a
library like Jackson to generate a string representation -- or look for
a custom OutputFormat that already does that for you.
I have a reducer that outputs pairs of the form<Text, MapWritable>. The
MapWritable is a map from Text to Long. The key and output values of my
reducer are declared like so:
job.setOutputKeyClass(Text.class);
job.setOutputValueClass(MapWritable.class);
The last few lines of my reducer look like this:
MapWritable value = new MapWritable();
for (Map.Entry<String, Long> entry : totalWordCount.entrySet())
value.put(new Text(entry.getKey()), new
LongWritable(entry.getValue()));
context.write(key, value);
In the debugger I can see that my MapWritable value is a map with the values
I expect it to contain. However, when I run the MapReduce process the
output literally looks like this:
key1 org.apache.hadoop.io.MapWritable@396cbd97
key2 org.apache.hadoop.io.MapWritable@17991de1
key3 org.apache.hadoop.io.MapWritable@18f63055
The values are descriptive stringifications of the MapWritable class without
any data. What I want instead are serializations of the maps that contain
their data. I thought the point of the MapWritable class was that it handled
serialization for you. Is there a step I'm missing?
Cheers,
Marcos
--
------------------------------------------------------------------------
Marcos Medrado Rubinelli
Estatística - Comparison Shopping - BuscaPé
