Hive likely wishes to format the data differently then Hadoop does. Hive re-uses what it can. I would diff the two .java files and find out for yourself :)
On Tue, Jun 12, 2012 at 5:20 PM, <richin.j...@nokia.com> wrote: > Hi Edward, > > Sorry, If I was not clear. My question is around difference between > DoubleWritable in hadoop and hive, other writables from hadoop works fine in > hive. > Hive.serde types are limited to Double, Byte, Short and Timestamp. > > I am using hive 0.8 > > Richin > > -----Original Message----- > From: ext Edward Capriolo [mailto:edlinuxg...@gmail.com] > Sent: Tuesday, June 12, 2012 5:12 PM > To: user@hive.apache.org > Subject: Re: hadoop.io.DoubleWritable v/s hive.serde2.io.DoubleWritable > > If you use Double or double hive will automatically convert. I would always > recommend the hive.serde types. > > Edward > > On Tue, Jun 12, 2012 at 4:56 PM, <richin.j...@nokia.com> wrote: >> Hi Guys, >> >> >> >> I am writing a UDF in hive to convert a double value to string, so the >> evaluate method of my UDF class looks like >> >> >> >> import org.apache.hadoop.hive.ql.exec.UDF; >> >> import org.apache.hadoop.io.Text; >> >> //import org.apache.hadoop.io.DoubleWritable; - does not work >> >> import org.apache.hadoop.hive.serde2.io.DoubleWritable; >> >> >> >> public Text evaluate(DoubleWritable d){ >> >> >> >> } >> >> >> >> When I looked at different UDF examples the Hadoop Writables seem to >> work fine in case of Text and IntWritable but for DoubleWritable I was >> getting an error. I figured out looking at couple of examples online >> that I should use the DoubleWritable from hive.serde2 package instead of >> hadoop.io package. >> >> >> >> Can someone please explain why this special case for DoubleWritable? >> >> >> >> Thanks, >> >> Richin
