Thanks, Julian! I think that's the last regression issue for 1.4.0 release from Drill side. With Calcite-833 patch, all the regression suites run successfully.
On Mon, Aug 10, 2015 at 12:20 PM, Julian Hyde <[email protected]> wrote: > Since 833 is a regression, yes, we can get it into the 1.4. I have added > some comments that I would like Sean to address. > > > On Aug 10, 2015, at 11:46 AM, Jinfeng Ni <[email protected]> wrote: > > > > Forgot to include the link to the JIRA of CALCITE-833 [1]. > > > > > > [1] https://issues.apache.org/jira/browse/CALCITE-833 > > > > On Mon, Aug 10, 2015 at 11:45 AM, Jinfeng Ni <[email protected]> > wrote: > > > >> The patch of DRILL-822 fixed the regressions we previously saw on Drill > >> side. > >> > >> There is one remaining regression in Drill's regression suite, for which > >> Sean has opened JIRA CALCITE-833 [1] and submitted a patch. Can we get > the > >> patch of CALCITE-833 merged into 1.4.0 release? > >> > >> > >> [1] https://github.com/hsuanyi/incubator-calcite/tree/CALCITE-833 > >> > >> > >> > >> > >> > >> On Sat, Aug 8, 2015 at 8:50 AM, Julian Hyde <[email protected]> wrote: > >> > >>> I have added 3 commits to master: Jesus' fix to 826, and Milinda's fix > >>> to 822, and Jinfeng's test case for 822. I think this fixes the main > >>> issues that were causing regression in Drill and Hive. I've created a > >>> new snapshot based on the latest master. > >>> > >>> Can those teams please re-test and let this list know how those tests > >>> went? > >>> > >>> A reminder that no other commits should go into master until the > >>> release is out. (Probably around 8/20.) I have been applying patches > >>> to my personal julianhyde/new-master branch[2], and you can consider > >>> this a preview of how the master branch will look right after the > >>> release. If you need the latest and greatest and can stand the > >>> occasional rebase, you're welcome to develop on that branch. > >>> > >>> Julian > >>> > >>> [1] > >>> > https://github.com/apache/incubator-calcite/commit/e827bf0a2e410e5613dcf407778e4f5735e75c1c > >>> > >>> [2] https://github.com/julianhyde/incubator-calcite/tree/new-master > >>> > >>> On Fri, Jul 31, 2015 at 2:29 PM, Julian Hyde <[email protected]> wrote: > >>>> +1 to let Jacques take vacation > >>>> > >>>> There's no urgency to get the release out. > >>>> > >>>> The only problem, and it's a minor one, is that master branch is > >>>> closed for check-ins until the release vote completes (August 22nd > >>>> even in the best case scenario). We won't have anywhere to integrate > >>>> changes that are to go into 1.5. > >>>> > >>>> One solution is to create a "branch-1.4" branch now, re-open master > >>>> for commits, and release from branch-1.4. But I'm not keen on that. > >>>> We'd have to start fixing bugs in two places, and the fixes will have > >>>> different hashes. > >>>> > >>>> I suggest creating a "future" branch and start committing changes for > >>>> the 1.5 release to that. Release 1.4 will be made from the "master" > >>>> branch. After the release, we will rebase "future" onto it. (And yes, > >>>> commit hashes may change during that rebase.) > >>>> > >>>> Julian > >>>> > >>>> > >>>> On Fri, Jul 31, 2015 at 11:10 AM, Ted Dunning <[email protected]> > >>> wrote: > >>>>> Speaking as a guy monitoring the mailing list while on vacation, I > >>> suggest > >>>>> that the delay is a good thing. > >>>>> > >>>>> Go away. Have fun. Relax. Chill. > >>>>> > >>>>> > >>>>> > >>>>> On Fri, Jul 31, 2015 at 9:50 AM, Jacques Nadeau <[email protected]> > >>> wrote: > >>>>> > >>>>>> Hey Guys, > >>>>>> > >>>>>> I'm leaving on vacation shortly. I was originally hoping to be all > >>> the way > >>>>>> through the release before the vacation. Then I was hoping for > being > >>> most > >>>>>> of the way through the release. It now looks like the first vote > >>> probably > >>>>>> won't have started before my vacation. I can run the release > process > >>>>>> during my vacation but I'd rather not (also a little worried about > >>> upload > >>>>>> speed). I suggest that we simply delay the release for a week and a > >>> half > >>>>>> (trying to get these issues solved during that time). Then I can > run > >>> the > >>>>>> release when I'm back to full speed on the 12th. > >>>>>> > >>>>>> Thoughts? > >>>>>> > >>>>>> thanks, > >>>>>> Jacques > >>>>>> > >>>>>> On Fri, Jul 31, 2015 at 9:42 AM, Jesus Camachorodriguez < > >>>>>> [email protected]> wrote: > >>>>>> > >>>>>>> We seem to have an assertion problem in line 394 in > RelFieldTrimmer. > >>>>>>> > >>>>>>> assert newProject.getClass() == project.getClass(); > >>>>>>> > >>>>>>> The new project is created with the provided factory, thus is of > >>> type > >>>>>>> HiveProject (this is correct); however, the project object is a > >>>>>>> LogicalProject. I need to track the plan changes to see when this > >>>>>>> LogicalProject is created. > >>>>>>> > >>>>>>> More to come... > >>>>>>> > >>>>>>> Thanks, > >>>>>>> Jesús > >>>>>>> > >>>>>>> > >>>>>>> > >>>>>>> On 7/31/15, 8:39 AM, "Jesus Camachorodriguez" > >>>>>>> <[email protected]> wrote: > >>>>>>> > >>>>>>>> Jacques, > >>>>>>>> > >>>>>>>> We run QA after the fix for CALCITE-821 went in, and I'm seeing 59 > >>> test > >>>>>>>> fails in Hive. I'm analyzing them right now... > >>>>>>>> > >>>>>>>> I will keep you posted. > >>>>>>>> > >>>>>>>> Thanks, > >>>>>>>> Jesús > >>>>>>>> > >>>>>>>> > >>>>>>>> > >>>>>>>> On 7/31/15, 3:36 AM, "Jacques Nadeau" <[email protected]> wrote: > >>>>>>>> > >>>>>>>>> Jesus, how are things looking after the new snapshot? > >>>>>>>>> On Jul 29, 2015 2:17 PM, "Julian Hyde" <[email protected]> wrote: > >>>>>>>>> > >>>>>>>>>> Yes, I just pushed a new snapshot. It's based on > >>>>>>>>>> > >>>>>>>>>> > >>>>>>>>>> > >>>>>>> > >>> > https://github.com/apache/incubator-calcite/commit/1580bd0fec62d836c9870 > >>>>>>>>>> e > >>>>>>>>>> dc339b0ab99d909cad > >>>>>>>>>> . > >>>>>>>>>> > >>>>>>>>>> On Wed, Jul 29, 2015 at 2:02 PM, Jesus Camachorodriguez > >>>>>>>>>> <[email protected]> wrote: > >>>>>>>>>>> Thanks for fixing it so quickly Julian. > >>>>>>>>>>> > >>>>>>>>>>> Sure, I'd like to do that. Is the new SNAPSHOT with the new > >>> fixes > >>>>>>>>>> already > >>>>>>>>>>> created? > >>>>>>>>>>> > >>>>>>>>>>> -- > >>>>>>>>>>> Jesús > >>>>>>>>>>> > >>>>>>>>>>> > >>>>>>>>>>> On 7/29/15, 10:55 PM, "Julian Hyde" <[email protected]> wrote: > >>>>>>>>>>> > >>>>>>>>>>>> The fix is checked in. Can you restart the Hive tests and let > >>> us > >>>>>> know > >>>>>>>>>>>> when they have passed? Or maybe you don’t think it necessary > >>> to run > >>>>>>>>>> the > >>>>>>>>>>>> Hive tests? > >>>>>>>>>>>> > >>>>>>>>>>>> Julian > >>>>>>>>>>>> > >>>>>>>>>>>> > >>>>>>>>>>>>> On Jul 29, 2015, at 11:10 AM, Jesus Camachorodriguez > >>>>>>>>>>>>> <[email protected]> wrote: > >>>>>>>>>>>>> > >>>>>>>>>>>>> Thanks a lot Julian. > >>>>>>>>>>>>> > >>>>>>>>>>>>> I'm checking it right now; I'll update the JIRA case when > >>> I'm > >>>>>> done. > >>>>>>>>>>>>> > >>>>>>>>>>>>> > >>>>>>>>>>>>> > >>>>>>>>>>>>> On 7/29/15, 7:43 PM, "Julian Hyde" <[email protected]> > >>> wrote: > >>>>>>>>>>>>> > >>>>>>>>>>>>>> Jesus, I have modified > >>>>>>>>>>>>>> https://issues.apache.org/jira/browse/CALCITE-821 with a > >>>>>>> suggested > >>>>>>>>>>>>>> patch. Can you see whether it works? > >>>>>>>>>>>>>> > >>>>>>>>>>>>>> On Wed, Jul 29, 2015 at 9:49 AM, Julian Hyde < > >>> [email protected]> > >>>>>>>>>> wrote: > >>>>>>>>>>>>>>> I'm looking into 821 also. You're right that it was > >>> caused by > >>>>>> the > >>>>>>>>>> edit > >>>>>>>>>>>>>>> to Frameworks.java in the fix to > >>>>>>>>>>>>>>> https://issues.apache.org/jira/browse/CALCITE-748. Not > >>> sure > >>>>>> why > >>>>>>>>>> that > >>>>>>>>>>>>>>> change was made, I'm pretty sure it was for a good reason. > >>>>>>>>>>>>>>> > >>>>>>>>>>>>>>> On Wed, Jul 29, 2015 at 12:58 AM, Jesus Camachorodriguez > >>>>>>>>>>>>>>> <[email protected]> wrote: > >>>>>>>>>>>>>>>> I've finally been able to trigger a Hive QA run. I have > >>> logged > >>>>>>>>>>>>>>>> CALCITE-821 > >>>>>>>>>>>>>>>> which is making all our tests fail. We will be checking > >>> into > >>>>>>>>>> that > >>>>>>>>>>>>>>>> today. > >>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>> Thanks, > >>>>>>>>>>>>>>>> Jesús > >>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>> On 7/29/15, 7:27 AM, "Jacques Nadeau" < > >>> [email protected]> > >>>>>>>>>> wrote: > >>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>> Ok. Will do. > >>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>> On Tue, Jul 28, 2015 at 10:15 PM, Julian Hyde > >>>>>>>>>> <[email protected]> > >>>>>>>>>>>>>>>>> wrote: > >>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>> Can you please take a look at 741, 743, Jacques. > >>> They're > >>>>>> best > >>>>>>>>>> done > >>>>>>>>>>>>>>>>>> while you're actually rolling a release. > >>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>> On Tue, Jul 28, 2015 at 9:49 PM, Jacques Nadeau > >>>>>>>>>>>>>>>>>> <[email protected]> > >>>>>>>>>>>>>>>>>> wrote: > >>>>>>>>>>>>>>>>>>> Julian, are you looking at 741/743 or should I take a > >>> look? > >>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>> On Tue, Jul 28, 2015 at 4:03 PM, Julian Hyde > >>>>>>>>>> <[email protected]> > >>>>>>>>>>>>>>>>>> wrote: > >>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>> I think you should include 811. It adds 3 new > >>> instances > >>>>>> of a > >>>>>>>>>> rule > >>>>>>>>>>>>>>>>>> but > >>>>>>>>>>>>>>>>>>>> does not change the existing one. > >>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>> I don't believe that Hive's test run has finished. > >>> Jesus, > >>>>>>>>>> can > >>>>>>>>>> you > >>>>>>>>>>>>>>>>>>>> report when you have some results? > >>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>> Still to do are release-related changes: release > >>> notes, > >>>>>> 741, > >>>>>>>>>> 743. > >>>>>>>>>>>>>>>>>> I > >>>>>>>>>>>>>>>>>>>> can do the web site after the RC. > >>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>> I was hoping to get Maryann's patch to 793 in, but I > >>> still > >>>>>>>>>> don't > >>>>>>>>>>>>>>>>>> have > >>>>>>>>>>>>>>>>>>>> a test case, so let's defer that one. > >>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>> Julian > >>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>> On Tue, Jul 28, 2015 at 3:23 PM, Jacques Nadeau > >>>>>>>>>>>>>>>>>> <[email protected]> > >>>>>>>>>>>>>>>>>>>> wrote: > >>>>>>>>>>>>>>>>>>>>> It seems like everyone is happy with the current > >>> release. > >>>>>>>>>> The > >>>>>>>>>>>>>>>>>> issue > >>>>>>>>>>>>>>>>>>>>> identified by Hive looks like it was a Hive fix > >>> rather > >>>>>> than > >>>>>>>>>> a > >>>>>>>>>>>>>>>>>> Calcite > >>>>>>>>>>>>>>>>>>>> one. > >>>>>>>>>>>>>>>>>>>>> I noticed that Jesus committed one thing after my > >>>>>> proposed > >>>>>>>>>>>>>>>>>> release. > >>>>>>>>>>>>>>>>>> Do > >>>>>>>>>>>>>>>>>>>>> people feel like that is absolutely required in this > >>>>>>>>>> release/the > >>>>>>>>>>>>>>>>>> risk > >>>>>>>>>>>>>>>>>>>>> involved? I'm fine including it but am initially > >>>>>> inclined > >>>>>>>>>> to > >>>>>>>>>>>>>>>>>> leave it > >>>>>>>>>>>>>>>>>>>> for > >>>>>>>>>>>>>>>>>>>>> the next release if it isn't pressing. The jira: > >>>>>>>>>>>>>>>>>>>>> https://issues.apache.org/jira/browse/CALCITE-811 > >>>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>>> Thoughts? > >>>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>>> On Mon, Jul 27, 2015 at 12:32 PM, Julian Hyde < > >>>>>>>>>> [email protected]> > >>>>>>>>>>>>>>>>>> wrote: > >>>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>>>> I¹ve tried the compile against Hive[1], fixed a > >>> minor > >>>>>>>>>> problem > >>>>>>>>>>>>>>>>>> (related > >>>>>>>>>>>>>>>>>>>> to > >>>>>>>>>>>>>>>>>>>>>> SortFactory), and now Jesus is going to run the > >>> Hive > >>>>>>>>>> suite. > >>>>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>>>> Jacques, If you¹re writing release notes you might > >>> find > >>>>>> my > >>>>>>>>>>>>>>>>>> relNotes[2] > >>>>>>>>>>>>>>>>>>>>>> script useful. > >>>>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>>>> Julian > >>>>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>>>> [1] > >>> https://issues.apache.org/jira/browse/HIVE-11383 > >>>>>>>>>>>>>>>>>>>>>> [2] > >>>>>>>>>>>>>>>>>> > >>>>>>> https://github.com/julianhyde/share/blob/master/tools/relNotes > >>>>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>>>> On Jul 24, 2015, at 7:35 PM, Jacques Nadeau > >>>>>>>>>>>>>>>>>> <[email protected]> > >>>>>>>>>>>>>>>>>> wrote: > >>>>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>>>>> I've pushed a 1.4.0-SNAPSHOT to the Apache > >>> Snapshot > >>>>>> Repo. > >>>>>>>>>>>>>>>>>> Framework > >>>>>>>>>>>>>>>>>>>>>>> owners, please try out this branch and report back > >>>>>>>>>> whether > >>>>>>>>>>>>>>>>>> you > >>>>>>>>>>>>>>>>>> see > >>>>>>>>>>>>>>>>>> any > >>>>>>>>>>>>>>>>>>>>>>> showstopping regressions. That way we can work > >>> out any > >>>>>>>>>> kinks > >>>>>>>>>>>>>>>>>> before > >>>>>>>>>>>>>>>>>>>> we > >>>>>>>>>>>>>>>>>>>>>> put > >>>>>>>>>>>>>>>>>>>>>>> this to an actual release vote. > >>>>>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>>>>> You can to the current commit by adding the > >>> following > >>>>>>>>>> maven > >>>>>>>>>>>>>>>>>> snapshot > >>>>>>>>>>>>>>>>>>>>>>> repository to your pom.xml: > >>>>>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>>>>> > >>>>>>>>>> https://repository.apache.org/content/repositories/snapshots/ > >>>>>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>>>>> The version of the artifacts to reference is > >>>>>>>>>> 1.4.0-SNAPSHOT. > >>>>>>>>>>>>>>>>>> The > >>>>>>>>>>>>>>>>>>>>>> artifacts > >>>>>>>>>>>>>>>>>>>>>>> are currently based on git commit > >>>>>>>>>>>>>>>>>>>>>> 5cb95e644f38da1a13d6def96eeec935b650289c. > >>>>>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>>>>> I'll target sending an actual release vote EOD > >>> Tuesday > >>>>>>>>>>>>>>>>>> unless we > >>>>>>>>>>>>>>>>>> find > >>>>>>>>>>>>>>>>>>>>>>> additional issues. > >>>>>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>>>>> Thanks, > >>>>>>>>>>>>>>>>>>>>>>> Jacques > >>>>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>>>> > >>>>>>>>>>>>>>>> > >>>>>>>>>>>>>> > >>>>>>>>>>>>> > >>>>>>>>>>>> > >>>>>>>>>>>> > >>>>>>>>>>> > >>>>>>>>>> > >>>>>>>> > >>>>>>> > >>>>>>> > >>>>>> > >>> > >> > >> > >
