The thin client is a newer feature. Some things like cli.print.header send output directly to the stream and the thrift client unfortunately only returns results over the thrift. It is a fair criticism, and sometimes annoying that output and error might go to a different place then the result data.
Like the open source world thinks get better little by little someone stepped up and made hive better by adding the header, Rome was not build in a day. Patches welcome. On Mon, May 28, 2012 at 1:10 PM, Stephen Sprague <[email protected]> wrote: > nah. no difference. okay. hive certainly shows immense potential but i > think i just have to fact the facts - its very immature at this time. > > On Fri, May 25, 2012 at 10:59 PM, Nitin Pawar <[email protected]> > wrote: >> >> try putting them in a file and execute as -f queryfile >> >> >> On Sat, May 26, 2012 at 7:51 AM, Stephen Sprague <[email protected]> >> wrote: >>> >>> Hi Hive Users, >>> >>> Please tell me if i am missing something. I prefer to run hive on a >>> thin client so i connect to a hive server by specifying a (remote) host and >>> port. (eg. $ hive -h <host> -p <port>) However, my sql results are not >>> coming back with headers - even though i explicitly request such. To wit, >>> see below: >>> >>> [namenode1:10000] hive> set hive.cli.print.header=true; >>> [namenode1:10000] hive> select * from sprague_test; >>> 1 sprague >>> 2 foo >>> 3 bar >>> >>> if i log on to the host where hadoop is running and run hive there w/o >>> specifying a host and port - i do get the headers. Moreover if i am on the >>> hadoop host and i run hive with -h and -p as localhost - i don't get the >>> headers! >>> >>> So whats up with hive.cli.print.header in remote mode? >>> >>> Thanks, >>> Stephen >> >> >> >> >> -- >> Nitin Pawar >> >
