Hi This has been fixed recently https://issues.apache.org/jira/browse/HIVE-4243
This used to be a problem with the way hive writes rows out. The ObjectInspectors sent out by hive’s filesink operator contains internal column names and not the names of the destination table. From the record reader’s perspective, the column names doesn’t really matter and column projections uses the column ids instead of names. Thanks Prasanth > On Oct 23, 2015, at 12:25 AM, patcharee <[email protected]> wrote: > > Hi > > I inserted a table from select (insert into table newtable select date, > hh, x, y from oldtable). After the insert the column names of the table > have been removed, see the output below when I use hive --orcfiledump > > - Type: struct<_col0:int,_col1:int,_col2:int,_col3:int> > > while it is supposed to be > > - Type: struct<date:int,hh:int,x:int,y:int> > > Any ideas how this happened and how I can fix it. Please suggest me. > > BR, > Patcharee >
