You can port the code to the new api yourself, perhaps? It isn't difficult
as much changes are superficial.

On Oct 6, 2010 9:52 PM, "Marc Sturlese" <[email protected]> wrote:


I'm working with hadoop 0.20.2 using the new API contained in the package:
org.apache.hadoop.mapreduce

I have noticed that MultipleInputs is under:
org.apache.hadoop.mapred
and when setting a path it asks for a JobConf:
addInputPath(JobConf conf, Path path, Class<? extends InputFormat>
inputFormatClass, Class<? extends Mapper> mapperClass)
but JobConf is deprecated in 0.20.2 (so all my jobs are using Job instead of
JobConf)

I have noticed too that in 0.21.0 there's no such problem as MultipleInputs
exist in org.apache.hadoop.mapreduce
and when setting a path it asks for a Job:
addInputPath(Job job, org.apache.hadoop.fs.Path path, Class<? extends
InputFormat> inputFormatClass, Class<? extends Mapper> mapperClass)

I was actually using 0.21.0 but had to downgrade to 0.20.2 as 0.21.0 can't
report progress in the reduce context (and this feature is a must for me):
https://issues.apache.org/jira/browse/MAPREDUCE-1905

So, wich would be the best way to use MultipleInputs? Should I change all my
code to use the org.apache.hadoop.mapred insetad of the
org.apache.hadoop.mapreduce classes?

I am really confused here, thanks in advance
--
View this message in context:
http://lucene.472066.n3.nabble.com/MultipleInputs-and-org-apache-hadoop-mapred-package-in-0-20-2-tp1643587p1643587.html
Sent from the Hadoop lucene-users mailing list archive at Nabble.com.

Reply via email to