Yeah, thanks Stephen. I dug~!
On Wed, Oct 23, 2013 at 5:58 PM, Stephen Sprague <[email protected]> wrote: > yeah. that works as expected. the schema drives the column list in the > select statement (not the hdfs file.) > > you'd have nulls if your schema had *more* columns than the hdfs file had > fields. > > you dig? > > > > > On Wed, Oct 23, 2013 at 4:53 PM, Xiu Guo <[email protected]> wrote: > >> We have a table called employee.dat with below contents: >> >> 1,ryan,d'souza,it,20000 >> 2,michael,fernandes,admin,25000 >> >> then in Hive, query: >> >> create table myTbl (a INT, b STRING) >> ROW FORMAT DELIMITED >> FIELDS TERMINATED BY ',' >> LINES TERMINATED BY '\n' >> TBLPROPERTIES ("serialization.null.format"="\\N"); >> >> LOAD DATA LOCAL INPATH "/.../employee.dat" overwrite into table myTbl; >> >> when do: >> select * from myTbl; >> >> the result is: >> >> 1 ryan >> 2 michael >> >> Is this correct? One of my teammate says if the dat file and table def >> has different number of columns, NULL values should be in the table. >> >> Can someone please confirm which one is expected behavior? >> >> Thanks, >> > >
