To clarify, HIVE-8435 introduced the regression. Turning that feature off fixes the issue. So we still need to fix that optimization to not produce this incorrect result.
On Fri, Jun 12, 2015 at 11:31 PM, Gautam <gautamkows...@gmail.com> wrote: > Found that turning off hive.optimize.remove.identity.project ( ref: > HIVE-8435 <https://issues.apache.org/jira/browse/HIVE-8435> ) fixes the > issue. > > This gives us a workaround, but dunno the performance degradation this > impacts yet. > > Thanks! > -Gautam. > > > On Fri, Jun 12, 2015 at 6:02 PM, Gautam <gautamkows...@gmail.com> wrote: > >> Done. https://issues.apache.org/jira/browse/HIVE-10996 >> >> On Fri, Jun 12, 2015 at 1:47 PM, Gopal Vijayaraghavan <gop...@apache.org> >> wrote: >> >>> Hi >>> >>> > Thanks for investigating.. Trying to locate the patch that fixes this >>> >between 1.1 and 2.0.0-SNAPSHOT. Any leads on what Jira this fix was part >>> >of? Or what part of the code the patch is likely to be on? >>> >>> git bisect is the only way usually to identify these things. >>> >>> But before you hunt into the patches I suggest trying combinations of >>> constant propogation, null-scan and identity projection remover >>> optimizations to see if there¹s a workaround in there. >>> >>> An explain of the query added to a new JIRA would be good, to continue >>> the >>> analysis. >>> >>> Cheers, >>> Gopal >>> >>> >>> >> >> >> -- >> "If you really want something in this life, you have to work for it. Now, >> quiet! They're about to announce the lottery numbers..." >> > > > > -- > "If you really want something in this life, you have to work for it. Now, > quiet! They're about to announce the lottery numbers..." > -- "If you really want something in this life, you have to work for it. Now, quiet! They're about to announce the lottery numbers..."
