So I think neither 6617 nor 6620 (which I extracted from 6617) were true regressions, they were separate bugs uncovered indeed by other changes I had made (but which didn't have good test coverage).
(For the curious, 6617 was a bug in TypeCoercion, which didn't handle most comparisons, and 6620 is a bug in the creation of VALUES, where the type for the result set of literal tuples was computed but never used in the construction of the set.) I have submitted PRs for both of them, hopefully these "regressions" are solved, but I have a feeling that more work will be necessary to fix other cases in TypeCoercion. This also explains why I have been complaining about Calcite RelNode programs missing casts: the coercion wasn't inserting the casts where it was actually meant to do it. Now I understand this part of the code much better and hopefully I can improve it. Hopefully this will unblock 1.38. Thank you for your patience and for the extra tests. Mihai ________________________________ From: Julian Hyde <jhyde.apa...@gmail.com> Sent: Monday, October 7, 2024 2:23 PM To: dev@calcite.apache.org <dev@calcite.apache.org> Subject: Re: [CANCEL] [VOTE] Release Apache Calcite 1.38.0 (release candidate 0) Mihai, Thanks for jumping on this. If you need more time then we could consider backing out the change that caused the problem. It’s in no one’s interests to hurry a fix. If necessary, we can make a release this week and then make another in two or three weeks. Julian > On Oct 7, 2024, at 2:02 PM, Mihai Budiu <mbu...@gmail.com> wrote: > > I am working on 6617, but I only have a reproduction for one of the two > reported problems. Was hoping it's quick, but I am discovering some > complications, I have discussed some of them in the JIRA case. > > Mihai > ________________________________ > From: Julian Hyde <jh...@apache.org> > Sent: Monday, October 7, 2024 1:55 PM > To: dev@calcite.apache.org <dev@calcite.apache.org> > Subject: [CANCEL] [VOTE] Release Apache Calcite 1.38.0 (release candidate 0) > > Thank you to Stamatis, Ruben and Mihai for discovering and logging cases > 6616 [1] and 6617 [2]. > > I agree that both are blockers, so I hereby cancel the vote. > > I am the cause of 6616 (I switched from JDK 8 to JDK 21 for the release > build) so I have assigned it to myself to fix before RC1. > > Can someone volunteer to fix 6617, and give an ETA for a fix? As soon as we > have a fix I create RC1 and start a vote. > > The main branch remains CLOSED. Let’s not introduce instability by fixing > things that are not broken. > > Thank you to everyone else who inspected the RC and voted. If you were > planning to check the release and vote, please continue to do so. It’s > possible that you will discover another issue, and the sooner we do that, > the sooner we shall have a good release. > > Julian > > > [1] https://issues.apache.org/jira/browse/CALCITE-6616 > [2] https://issues.apache.org/jira/browse/CALCITE-6617 > > > On Oct 7, 2024, at 3:45 AM, Ruben Q L <rube...@gmail.com> wrote: > > Thanks Julian for preparing this release. > > - Release notes: ok (minor detail: IIANM this release comes 5 months after > 1.37, not 6) > - Checksum: ok > - Signature: ok > - Diff source release and git repository: ok > - Build + tests: ok > - Calcite-based application test suite: KO > It seems there is a regression (kudos to Claude Brisson for discovering > it), the issue has been logged here: > https://issues.apache.org/jira/browse/CALCITE-6617 > > -1 rubenql (PMC member) > > Best, > Ruben > > > > On Mon, Oct 7, 2024 at 10:05 AM Stamatis Zampetakis <zabe...@apache.org> > wrote: > > -1 zabetak (PMC member) > > I am changing my vote cause it seems that maven released artifacts are > unusable by projects using JDK < 21. I logged CALCITE-6616 [1] for > tracking this further. > > [1] https://issues.apache.org/jira/browse/CALCITE-6616 > > On Mon, Oct 7, 2024 at 9:28 AM Stamatis Zampetakis <zabe...@apache.org> > wrote: > > > Ubuntu 20.04.6 LTS, jdk1.8.0_261, Gradle wrapper, Gradle 8.7 > > * Checked signatures and checksums OK > * Checked diff between repo and artifacts (diff -qr calcite-src > calcite-git) OK (LICENSE file between the release sources and git tag > contains small differences) > * Went over release note OK > * Checked README has correct version and build instructions OK > * Checked NOTICE OK > * Checked LICENSE OK (Changes from CALCITE-6098 are not present) > * All source files have ASF headers (grep -RiL "Licensed to the > Apache Software Foundation") OK > * No unexpected binary files OK (find . -type f -exec file {} \; | > grep -v text) > * Built from source artifacts and run tests (gradle build) OK > * Built from git tag and run tests (./gradlew build) OK > * Checked LICENSE, NOTICE, for calcite-core-1.38.0.jar in nexus OK > > +1 zabetak (PMC member) > > Best, > Stamatis > > > > > On Sat, Oct 5, 2024 at 10:49 PM Julian Hyde <jh...@apache.org> wrote: > > > Thanks for that research, Francis. After the release I'll fix > release/build.gradle.kts (which generates the vote email) to generate > the correct link. I'll also update the 'Your vote should contain the > line' part of the email. > > On Sat, Oct 5, 2024 at 2:00 AM Francis Chuang < > > francischu...@apache.org> wrote: > > > I think the tag was already pushed previously, because I am pretty > > sure > > I saw the tag on GitHub when I voted earlier in the day. > > The problem is that there's a bug in GitBox that redirects the tag to > > https://github.com/apache/calcite/commit/refs/tags/calcite-1.38.0-rc0 > > when the correct link is > https://github.com/apache/calcite/tree/calcite-1.38.0-rc0 > > On 5/10/2024 4:52 pm, Julian Hyde wrote: > > The Tag's web link is invalid. > > > Sorry, I forgot to push the tag. It should work now. > > (Please continue to vote. My mistake does not invalidate the vote.) > > On Fri, Oct 4, 2024 at 9:26 PM Xiong Duan <xi...@apache.org> > > wrote: > > > +1 Xiong Duan(Committer) > > The Tag's web link is invalid. > > Julian Hyde <jh...@apache.org> 于2024年10月5日周六 11:36写道： > > > Hi all, > > I have created a build for Apache Calcite 1.38.0, release > candidate 0. > > Thanks to everyone who has contributed to this release. > > You can read the release notes here: > > https://github.com/apache/calcite/blob/calcite-1.38.0-rc0/site/_docs/history.md > > > The commit to be voted upon: > > https://gitbox.apache.org/repos/asf?p=calcite.git;a=commit;h=d3d309f5f05b1d28e59440d3f4d69d12e6d6efb9 > > > Its hash is d3d309f5f05b1d28e59440d3f4d69d12e6d6efb9 > > Tag: > > https://gitbox.apache.org/repos/asf?p=calcite.git;a=tag;h=refs/tags/calcite-1.38.0-rc0 > > > The artifacts to be voted on are located here: > > https://dist.apache.org/repos/dist/dev/calcite/apache-calcite-1.38.0-rc0 > > (revision 72135) > > The hashes of the artifacts are as follows: > > bba61cfd946898250c001b514902bcb84b6be8357c2be491f14fb3dc3cd3702d329976fb82d536e673c63da00bcc4d2a4bdbd9db3c31c7d41fb95211c0ef84bb > > *apache-calcite-1.38.0-src.tar.gz > > A staged Maven repository is available for review at: > > https://repository.apache.org/content/repositories/orgapachecalcite-1233/org/apache/calcite/ > > > Release artifacts are signed with the following key: > https://people.apache.org/keys/committer/jhyde.asc > https://www.apache.org/dist/calcite/KEYS > > To create the jars and test Apache Calcite: "gradle build" > (requires an appropriate Gradle/JDK installation) > > Please vote on releasing this package as Apache Calcite 1.38.0. > Everyone is welcome to vote. Your vote should contain the line > 'code name (role)', where > * 'code' is +1, 0 or -1 as below; > * 'name' is your ASF id if you have one, else your real name; > * 'role' is 'community', 'committer' or 'PMC member'. > > [ ] +1 Release this package as Apache Calcite 1.38.0 > [ ] 0 I don't feel strongly about it, but I'm okay with the > > release > > [ ] -1 Do not release this package because... > > The vote is open for the next 96 hours (closing at 0400 UTC on > > Wed Oct > > 9th) and passes if a majority of at least three +1 votes by PMC > members are cast. > > Here is my vote: > > +1 jhyde (PMC member) > > Julian Hyde (release manager)
