+1 binding Ran with provided script. Python and Helm all look good.
Thanks, Yong > On Nov 12, 2026, at 1:04 PM, Yufei Gu <[email protected]> wrote: > > +1 binding > > I verified: > > - Tag/commit and source tarball contents: OK > - Signatures and checksums, including Maven artifacts: OK > - Build from the tag, license report, and RAT: OK > - LICENSE/NOTICE checks, including Spark JARs: OK > - Tag-built server and binary distribution smoke tests: OK > - Python CLI: OK > - Helm provenance, lint, and template rendering: OK > - Run bin/server, bin/admin from the binary distribution: OK > > Yufei > > >> On Thu, Nov 12, 2026 at 7:29 AM Alex Dutra via dev <[email protected]> >> wrote: >> >> +1 (binding) >> >> I verified the following for Apache Polaris 1.9.0-rc1: >> >> - Automated checks (`verify-release.sh`): OK (byte-for-byte comparisons >> off) >> - License report validation (`./gradlew generateLicenseReport`): OK >> - CHANGELOG/version.txt: OK >> - LICENSE/NOTICE manual review: OK >> - Spark plugin JARs: OK >> - Smoke tests: >> - Binary distribution: OK >> - Source distribution: OK >> - Helm chart: OK >> >> Thanks, >> Alex >> >> >> >>> On Thu, Nov 12, 2026 at 3:16 PM Robert Stupp <[email protected]> wrote: >>> >>> +1 (binding) >>> >>> Followed the release-verification guide [1] using the >> release-verification >>> script. >>> >>> I ran that script with byte-for-byte comparisons enabled and investigated >>> the reported differences. >>> >>> The differences fall into these groups: >>> >>> * Javadoc JARs: My local built and the staged artifacts have differences, >>> as the release-workflow ran [1] with Temurin 21.0.13+9, while my local >>> build used 21.0.12. >>> The differences are limited to two JavaScript resources bundled >>> with the JDK’s standard doclet, consistent with different JDK builds. >>> The remaining Javadoc content compares equal. >>> The `.module` differences are a downstream effect: only the sizes and >>> checksums of the corresponding Javadoc artifacts differ. >>> >>> * Binary distributions: the differences are confined to two embedded >> Quarkus >>> `generated-bytecode.jar` files. >>> These contain an ambient build-locale value and differences in >>> config-mapping >>> registration order. >>> The config-generation fixes are proposed upstream in >>> [Quarkus-58360] and [Quarkus-58361), the locales fix is in [6056]. >>> >>> * License-report ZIPs: I excluded these from the detailed comparison, >> given >>> the >>> planned jk1 plugin replacement covered in the License/SBOM discussion >> [2]. >>> >>> All 188 other Maven JARs and all 87 POMs compared byte-for-byte equal. >>> >>> For future releases, [6079] pins the release tool versions, includes the >>> selected JDK version/vendor in compilation and documentation task inputs, >>> and publishes signed build-environment evidence to help diagnose >>> reproduction differences. >>> >>> [1] >>> >> https://github.com/apache/polaris/actions/runs/42968493288/job/130351412090#step:5:28 >>> [Quarkus-56360] https://github.com/quarkusio/quarkus/pull/58360 >>> [Quarkus-56361] https://github.com/quarkusio/quarkus/pull/58361 >>> [2] >> https://lists.apache.org/thread.html/cm3wb15vg7ws4j98bpq1myt5n18j635h >>> [6056] https://github.com/apache/polaris/pull/6056 >>> [6079] https://github.com/apache/polaris/pull/6079 >>> >>> >>> >>> On Thu, Nov 12, 2026 at 12:32 PM Ajantha Bhat <[email protected]> >> wrote: >>> >>>> +1 (non-binding) >>>> >>>> Verified: >>>> >>>> - Signatures & Checksums >>>> - License & Notice >>>> - Build from source >>>> >>>> - Ajantha >>>> >>>> On Thu, Nov 12, 2026 at 10:22 AM Xiaodong Deng via dev < >>>> [email protected]> wrote: >>>> >>>>> +1 (non-binding) >>>>> >>>>> - Ran tools/verify-release/verify-release.sh from the rc1 tag: >>>>> LICENSE/NOTICE and version.txt are in place, and the signatures and >>>> SHA-512 >>>>> checksums of the source tarball all verify. >>>>> - Python client: the wheel's signature and checksum look good. >>>>> - Used the 1.9.0 client for an application I have been building with >>>>> 1.8.0, and confirmed 1.9.0 could work well there too. >>>>> >>>>> One observation (not a blocker): >>>>> As the CHANGELOG notes, renaming a table into another namespace now >>>>> returns 403 unless its files lie within the destination namespace's >>>>> location (ALLOW_UNSTRUCTURED_TABLE_LOCATION=false by default). >>>>> Downstream code that moves tables across namespaces may need >> adjusting. >>>>> This became a little surprise for my application during the testing. >>>>> I think it may be worth highlighting in the communications when the >>>>> release is finalized & announced. >>>>> >>>>> Last but not least: thanks for preparing the release! >>>>> >>>>> >>>>> XD >>>>> >>>>> >>>>> On 2026/11/11 18:39:34 Prithvi S wrote: >>>>>> +1 (non-binding) >>>>>> >>>>>> Verified: >>>>>> * Signatures >>>>>> * Checksums >>>>>> * LICENSE / NOTICE >>>>>> * Build from source >>>>>> * No unexpected binaries in the source archive >>>>>> * polaris-bin starts with in-memory persistence; catalog, >> namespace and >>>>>> table operations work as expected >>>>>> * Python CLI wheel installs and runs successfully >>>>>> >>>>>> Thanks, JB, for driving the release. >>>>>> >>>>>> Thanks, >>>>>> Prithvi >>>>>> >>>>>> On Wed, Nov 11, 2026 at 11:12 PM vignesh a <[email protected] >>> >>>>> wrote: >>>>>> >>>>>>> +1 (non-binding) >>>>>>> >>>>>>> * Built from source >>>>>>> * Signatures and checksums for the source tarball, polaris-bin, >> and >>>>> Helm >>>>>>> chart >>>>>>> * LICENSE and NOTICE in the source tarball and polaris-bin >>>>>>> * No unexpected binaries in the source tarball >>>>>>> >>>>>>> Thanks JB for driving the release. >>>>>>> >>>>>>> Thanks, >>>>>>> Vignesh >>>>>>> >>>>>> >>>>> >>>> >>
