+1 (non-binding)
[x] Checksums and PGP signatures are valid [x] LICENSE and NOTICE files are correct [x] All files have ASF license headers where appropriate [x] No unexpected binary files [x] Source tarball matches the Git tag [x] Rebuilt tarballs match the staged SHA512 checksums [x] Can compile from source successfully [x] Docker Compose bundle deploys successfully with the published images What I checked, on macOS arm64 with Docker 28: - Both .asc signatures verify against D4F761FF20CF28580A5A9D1D80BBC55290D0C350 from the KEYS file, and both SHA-512 checksums match. - The extracted source tarball is identical to tag v1.3.0-incubating-rc2 (177ed20102d42813bc9a15bb1781d6370ddc822f), apart from .git. - I rebuilt both tarballs from the tag in an ubuntu:24.04 container (GNU tar 1.35, gzip 1.12). Both SHA-512s match the staged ones. - Apache RAT 0.16.1 and skywalking-eyes 0.8.0 (with .licenserc.yaml) found 0 files with invalid headers. Every file RAT flags is a license text, small config or test data, generated code, or third-party code attributed in LICENSE. - The binary files are images, tutorial GIFs, and the sample audio and video the contributor made for the Hugging Face operator. There are no jars, class files or other compiled code. - Backend: sbt compile, Test/compile and dist for all 7 services passed on JDK 17, with jOOQ codegen run against a fresh Postgres loaded with sql/*.sql. - Frontend: yarn install --immutable and yarn build:ci passed on Node 24. - Docker Compose: ran "docker compose --profile examples up" on the unmodified bundle (linux/arm64 images). All services came up healthy, the example dataset and workflow loaded, and I logged in as texera. I created a local computing unit and ran "[Example] Machine Learning on Iris Dataset" end to end, including the Python sklearn operators. - I also confirmed the disclosed core dump in the arm64 coordinator image (/qemu_ldconfig.real_20261110-064228_3633.core, 16.8 MB). Since the source release isn't affected, I don't consider it blocking. One non-blocking note on reproducibility: the instructions say "GNU tar", but the version matters. Debian bookworm's GNU tar 1.34 writes "0000000" into the devmajor/devminor header fields of regular files, and 1.35 leaves them empty. So a rebuild on bookworm gives different bytes: the file contents are identical but the tar headers differ. It would help to say "GNU tar 1.35+" or to suggest running the script in ubuntu:24.04. On 2026-11-16 01:22 UTC Meng Wang via general wrote: > Hi Incubator Community, > > This is a call for a vote on releasing Apache Texera (Incubating) > 1.3.0-incubating RC2. > > == PPMC Vote Result == > This release candidate has been approved by the Apache Texera PPMC. > > PPMC vote thread: > https://lists.apache.org/thread/s9w8dcv1xco820mg26okxvxhtd1jyn4m > > PPMC vote result: > https://lists.apache.org/thread/5jps9wmccdxd2oht0f6wj7vj4lnf1k0w > > == Major Changes Since Apache Texera 1.2.0-incubating == > Full diff: > https://github.com/apache/texera/compare/v1.2.0-incubating...v1.3.0-incubating-rc2 > > Major changes include: > Deployment and Storage > - Replaced MinIO with RustFS as the default object store. > - Made Lakekeeper's REST catalog the default Iceberg catalog. > - Allowed Kubernetes deployments to use external S3 storage. > > AI Integration > - Added a Hugging Face inference operator with text, image and audio tasks. > > Workflow Engine and Operators > - Added loop operators. > - Added UI parameters for Python UDFs. > - Enabled persistent Python virtual environments for custom packages. > - Improved input validation in visualization and ML operators. > > Datasets > - Added dataset contributor metadata, linked to user accounts. > - Improved resumable dataset uploads and cleanup of stale uploads. > > User Experience > - Added card views for workflows and datasets. > - Added a user feedback dashboard for administrators. > - Added email verification at sign-up. > > Release and Quality > - Made the release tarballs and container images reproducible. > - Replaced assets of unknown origin and updated LICENSE and NOTICE. > - Substantially expanded unit test coverage across the codebase. > - Added bin/local-dev.sh for running the full stack locally. > > == Release Candidate Artifacts == > https://dist.apache.org/repos/dist/dev/incubator/texera/1.3.0-incubating-RC2/ > > The directory contains: > - Source tarball (.tar.gz) with a GPG signature (.asc) and SHA-512 > checksum (.sha512) > - Docker Compose deployment bundle with a GPG signature and SHA-512 checksum > > == Container Images == > The following images are available on Docker Hub: > > apache/texera-dashboard-service:1.3.0-incubating-rc2 > apache/texera-workflow-execution-coordinator:1.3.0-incubating-rc2 > apache/texera-workflow-execution-runner:1.3.0-incubating-rc2 > apache/texera-workflow-compiling-service:1.3.0-incubating-rc2 > apache/texera-file-service:1.3.0-incubating-rc2 > apache/texera-config-service:1.3.0-incubating-rc2 > apache/texera-access-control-service:1.3.0-incubating-rc2 > apache/texera-agent-service:1.3.0-incubating-rc2 > apache/texera-workflow-computing-unit-managing-service:1.3.0-incubating-rc2 > apache/texera-notebook-migration-service:1.3.0-incubating-rc2 > apache/texera-jupyter:1.3.0-incubating-rc2 > > These images are built from the source tarball included in this release. > The Dockerfiles are included in the source for audit and verification. > > Note: I found that the linux/arm64 texera-workflow-execution-coordinator > image contains an extra core dump file from an ldconfig crash under QEMU > during the CI build. ldconfig was rerun successfully, so the image works > normally and otherwise matches a rebuild from the tag, but the extra file > makes it not bit-for-bit reproducible. The source release is not affected. > > == Git Tag == > https://github.com/apache/texera/releases/tag/v1.3.0-incubating-rc2 > Commit: 177ed20102d42813bc9a15bb1781d6370ddc822f > > == Reproducing the Artifacts == > Both tarballs are reproducible. To rebuild them from the tag (needs GNU tar > and GNU gzip; macOS's built-in gzip produces different bytes, so on macOS > run this in a Linux container) and compare against the staged checksums: > > git clone https://github.com/apache/texera.git && cd texera > git checkout v1.3.0-incubating-rc2 > ./.github/scripts/create-release-tarballs.sh v1.3.0-incubating-rc2 \ > 1.3.0-incubating docker.io/apache 1.3.0-incubating-rc2 /tmp/texera-rc > sha512sum /tmp/texera-rc/*.tar.gz > > == Keys == > The release was signed with GPG key > [D4F761FF20CF28580A5A9D1D80BBC55290D0C350] ([email protected]) > KEYS file: https://downloads.apache.org/incubator/texera/KEYS > > == Vote == > The vote will remain open for at least 72 hours. > The vote passes with at least three binding +1 votes and more +1 than -1 > votes. > > [ ] +1 Approve the release > [ ] 0 No opinion > [ ] -1 Disapprove the release (please provide a reason) > > == Checklist == > Please include the checklist below in your reply and mark the checks > you performed with an x. > > [ ] Checksums and PGP signatures are valid > [ ] LICENSE and NOTICE files are correct > [ ] All files have ASF license headers where appropriate > [ ] No unexpected binary files > [ ] Source tarball matches the Git tag > [ ] Rebuilt tarballs match the staged SHA512 checksums > [ ] Can compile from source successfully > [ ] Docker Compose bundle deploys successfully with the published images > > Thank you for reviewing and voting on this release candidate. > > Best, > Meng Wang > Release Manager for Apache Texera v1.3 > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > > -- Best Regards, Kunwoo (Chris) Park ---------------------------------------------------------------------------- Kunwoo (Chris) Park He/Him/His Ph.D. Student Donald Bren School of Information & Computer Sciences 2059 Bren Hall (949) 413-5324 ----------------------------------------------------------------------------
