June 2, 2026

Monthly sync meeting summary for June 2, 2026

Texera Apache Incubation – Monthly Meeting Minutes

Date: June 2, 2026
Participants:

Texera Committers: Xuan Gu, Ali Risheh, Jiadong Bai, Chris Park, Yicong Huang, Xiaozhen Liu, Meng Wang, Xinyuan Lin
Texera Contributors: Carlos Ernesto Alvarez Berumen, Grace Chia
Apache Incubating Mentors: Ian Maxon
Guests: Eugene Gu, Kary Zheng, Kyle Dang, Suryaa

The dkNET-AI · Apache Texera Agent Hackathon — the project’s first hackathon — drew 35 submissions and 10 awardees, and brought new developers into the community.

“Supporting Our AI Overlords (SAO): AI Agents and Data Systems” was presented at the ACM Conference on AI and Agentic Systems (CAIS) on May 26, 2026 in San Jose, CA. Jiadong Bai represented the project.

The project shipped its first Apache release, v1.1.0-incubating (released May 17, 2026), a major milestone since v1.0.0 (April 4, 2025).
The release represents 800+ commits and over 1M changed lines of code. A LinkedIn announcement was published to help promote it.
A large share of the chore/CI work was governance and compliance:
- Trimming the DISCLAIMER to the ASF template.
- Per-module LICENSE-binary/NOTICE files.
- A strict license-drift checker.
- ASF branch-protection rulesets.
- Release-candidate (RC) bundling tooling.

Major release (e.g., v2.0): change of position/focus, significant large feature, or marketing/branding update.
Minor release (e.g., v1.2): new features, refactoring, bug fixes.
Patch release (e.g., v1.1.1): urgent bug fix; used only when absolutely needed.

Branch cuts follow a quarterly cadence with a targeted release roughly two weeks later:
- v1.1 — branch cut March 1, 2026 (v1.1.0-incubating released May 17, 2026)
- v1.2 — branch cut June 1, 2026, targeted June 15, 2026
- v1.3 — branch cut September 1, 2026, targeted September 15, 2026
- v1.4 — branch cut December 1, 2026, targeted December 15, 2026
- v1.5 and v1.6 — projected for 2027
Major versions are bumped strategically; patch versions only when absolutely needed.

Coordinates the release timeline and branch cut; prepares RCs, source packages, tags, checksums, and signatures; coordinates the release vote and the incubation release vote; publishes approved artifacts; updates the website/download links and announcements; and keeps the community informed.
One release manager owns one release line (e.g., the RM of v1.2 manages v1.2.0-incubating, v1.2.1-incubating, …). By default only one release is needed per line. Watch for the [CALL FOR VOLUNTEER] email on dev@texera.apache.org.
Thanks to Xuan Gu for volunteering as RM of the v1.2 line, and Jiadong Bai as RM of the v1.1 line.

To keep work organized, different kinds of conversations belong in different places:

GitHub Discussions — “Do we do it? Which option?” Decide which feature or direction to take, which UI/UX to provide, and which design option to choose among alternatives.
GitHub Issues — “When/where (release) to do it?” Define the concrete task, bug, or feature (often summarized from a discussion), decide priority and targeted release, and coordinate with other work.
GitHub PRs — “How do we implement it?” Verify the implementation is correct and has enough verification (tests, screenshots, videos, benchmarks).

Discussions are synced to dev@texera.apache.org.

Everyone is encouraged to review any PRs; contributors with more context should be requested as reviewers.
AI reviews (local agents such as Claude Code and Codex, plus GitHub Copilot) are encouraged to add perspective, but should not replace human reviews — they provide an extra pair of eyes.
Review and merge life cycle: author requests review → reviewer comments / requests changes → author addresses and re-requests review → reviewer approves (with non-blocking comments) → author resolves non-blocking comments and applies final touches → merger merges.
Roles:
- Author: make the code change, create the PR and link the issue, add tests that prove the change works, include screenshots/gifs for UI PRs, and request reviewers. Authors should do the heavy lifting to reduce reviewer load.
- Reviewer: check that the PR matches its issue/discussion, assess code quality and test coverage, and optionally test to verify. Reviewing any PR you’re interested in is welcome; even lightweight feedback is valuable.
Who merges: if the author is a committer, the author merges; if the author is a contributor, a reviewer merges; in rare cases where both are unavailable, a third committer may merge.

Testing is especially significant for AI-generated code, and should be automated as much as possible in CI (unit tests — the current focus — plus E2E tests and benchmarks).
Unit test coverage is reported via Codecov; the repository started at roughly 42% coverage.
Filling the coverage gap helps avoid regressions (breaking old features, reintroducing bugs, or unnoticed behavior changes) and reduces per-PR review and per-release manual bug-bash workload.
Approach:
- Encourage test-only PRs to add tests.
- For new feature/task PRs, aim for 100% coverage as guidance (not a hard requirement; the committer has final say). 0% coverage on changed code indicates it is untested and should be rejected.
- For bug-fix PRs, add tests to confirm the fix and prevent recurrence.
Goal: reach 90% test coverage by the end of summer.
New CI/CD infrastructure: GitHub merge queue, label-based optional pre-merge CI plus full post-merge CI, label-based auto-backport workflows for release branches, and a contributor-facing comment-command bot.

Apache name approved: the “Apache Texera” name was approved by the ASF (PODLINGNAMESEARCH-253).
Agent Service: a new agent-service was added to manage LLM agents (#4495), enabling the Texera Agent to edit and execute workflows (#4540) and to support more default operator types (#4619).
Python Virtual Environment (PVE): creating environments (#4484), installing/uninstalling user packages (#4902, #5035), executing Python UDFs inside PVEs (#5069), a UI parameter form (#5043), and a Kubernetes gateway config (#5138).
New visualization operators: Radar, Wind rose, Parallel coordinates, ECDF, Carpet, Ternary contour, and WebGL polar charts.
Engine (Amber): operator reconfiguration re-enabled (#4220), region restart support (#4441), state materialization across regions (#4490), jump-to-operator support (#4444), and Iceberg REST catalog support for result storage (#4272).
Frontend: upgraded Angular 17 → 21 and migrated from Karma to Vitest.
Backend: moved to Java 17 LTS, and Akka was renamed to Pekko (the ASF-licensed fork).
New data-source operators: File Lister and File Scan From Input.
Quality and testing: a deliberate stabilization phase — 111 of 401 commits (over a quarter) were test commits — paired with full Codecov integration and a coverage badge.

Ongoing efforts the team will continue: