Absolutely agree the fix would be good to get in. My comments come from the fact that we tried to start the 1.9 release (late last year I think) after the direct memory feature got in and it never completed (as there is always a good reason to hold the release). My main fear is that if we wait for something and it turns into another substantial delay before a release. Ryan, what do you think about proposing a hard deadline?
On Thu, Sep 8, 2016 at 7:57 PM, Robert Kruszewski <[email protected]> wrote: > When would the fix be ready by to consider it for release in 1.9.0? This > affects correctness on current master of spark and would be good to get a > release with it fixed so that people aren’t using potentially bad versions. > > - Robert > > On 9/8/16, 10:44 PM, "Jacques Nadeau" <[email protected]> wrote: > > A non-binding +1 from me on releasing sooner/more often. > > On Thu, Sep 8, 2016 at 5:44 PM, Ryan Blue <[email protected]> > wrote: > > > Hey everyone, > > > > I'd like to put together a release candidate for 1.9.0. The other > issues > > are done, but the sort order min/max issue, PARQUET-686 is still > open. > > > > I'm okay releasing 1.9.0 without fixing that issue since it's been > so long > > since our last release. It would also be nice to do releases as > necessary, > > so we can always do a release to fix PARQUET-686 when the patch for > it is > > ready. Is there anyone that thinks we should definitely get this > into the > > 1.9.0 release? > > > > Thanks, > > > > rb > > > > -- > > Ryan Blue > > Software Engineer > > Netflix > > > >
