Its hard to tell for sure from that snippet if you're altering the actual job configuration or a new, disconnected configuration object. You need to do the former to have it apply.
On Thu, Jul 18, 2013 at 11:52 PM, Botelho, Andrew <[email protected]> wrote: > I am doing exactly what this website tells: > http://cloudfront.blogspot.com/2012/06/how-to-change-default-key-value.html > > But it isn’t changing anything. > > > > Andrew > > > > From: Ravi Kiran [mailto:[email protected]] > Sent: Thursday, July 18, 2013 2:16 PM > > > To: [email protected] > Subject: Re: Make job output be a comma separated file > > > > Hi Andrew, > > > > Can you please tell me which version of Hadoop you use.. I noticed that in > Hadoop 1.0.4 , the class > org.apache.hadoop.mapreduce.lib.output.TextOutputFormat is looking for > mapred.textoutputformat.separator . > > Regards > > Ravi M > > > > On Thu, Jul 18, 2013 at 11:32 PM, Botelho, Andrew <[email protected]> > wrote: > > I believe that mapred.textoutputformat.separator is from the old API, but > now the field is mapreduce.output.textoutputformat.separator in the new API. > > So I ran this code in my driver class, but it is making no difference: > > > > Configuration conf = new Configuration(); > > conf.set("mapreduce.output.textoutputformat.separator", ","); > > > > Am I changing the field right? > > > > Thanks, > > Andrew > > > > From: Ravi Kiran [mailto:[email protected]] > Sent: Thursday, July 18, 2013 1:45 PM > To: [email protected] > Subject: Re: Make job output be a comma separated file > > > > Hi Andrew, > > > > You can pass change the default keyValueSeparator of the output format > from a "\t" to a "," by > setting the following property mapred.textoutputformat.separator to > Configuration of the job. > > You will face difficulties if this output is an input to another job as > you wouldn't know what part of the row data is a key and what is the value. > > > > Regards > > Ravi M. > > > > On Thu, Jul 18, 2013 at 10:46 PM, Botelho, Andrew <[email protected]> > wrote: > > What is the best way to make the output of my Hadoop job be comma separated? > Basically, how can I have the keys and values be separated by a comma? > > My keys are Text objects, and some of them have actual commas within the > field. Will this matter? > > > > Thanks, > > > > Andrew > > > > -- Harsh J
