FYI, HBASE-20276 has now been committed to all impacted branches. If folks have a chance to review the proposed release note, that'd be grand.
On Fri, Apr 6, 2018 at 2:55 PM, Josh Elser <[email protected]> wrote: > +1 ditto -- this one is better to revert and ask forgiveness on from users, > IMO. > > > On 4/3/18 3:27 PM, Apekshit Sharma wrote: >> >> +1 on going back to old behavior, i.e. returning values, in branch-1 and >> 2.0 release. >> >> bq. Open question: should we also revert this change in branch-1.4, even >> though >> it means the shell will behave very differently between maintenance >> releases? >> Earlier, i was inclining towards reverting the behavior in next 1.4.x too, >> but not so much now. Probably a sooner 1.5 is better idea? >> >> On Tue, Apr 3, 2018 at 6:50 AM, Sean Busbey <[email protected]> wrote: >> >>> Hi folks! >>> >>> Wanted to bring your attention to HBASE-20276 "[shell] confirm shell >>> REPL change and document". >>> >>> tl;dr: there's a change currently in 1.4.{0..3} and 2.0.0-* that makes >>> us not return values from any shell commands. current consensus is >>> that this fundamentally breaks the shell and we should revert it. At >>> least one person (me) believes that we should not treat releases with >>> this behavior as our stable / recommend release line. >>> >>> Patch shortly going in for branch-1 and newer that reverts the behavior. >>> >>> Open question: should we also revert this change in branch-1.4, even >>> though it means the shell will behave very differently between >>> maintenance releases? >>> >> >> >> >
