Urgent.News

What's breaking now, across thousands of outlets.

AI

PostgreSQL 19 Slips to November as AI Code Review Finds Bugs Faster Than Humans Can Fix Them

PostgreSQL 19 Slips to November as AI Code Review Finds Bugs Faster Than Humans Can Fix Them The PostgreSQL Global Development Group has delayed PostgreSQL 19, pushing general availability from the original late-September/early-October target to around November 19, 2026 — and the culprit, remarkably, is not sloppy engineering. It is the opposite: AI-assisted code review has become so effective…

The PostgreSQL Global Development Group has pushed back the general availability of PostgreSQL 19 from late September/early October to November 19, 2026. This delay is not due to any shortcomings in the code itself but rather to the effectiveness of AI-assisted code review. According to Peter Eisentraut, a leading PostgreSQL committer and VP at EDB, the release team had to revert between eight and ten significant features to PostgreSQL 20.

These reverts were not due to architectural issues but rather because AI review surfaced a large number of defect reports that needed to be addressed. During the beta period, AI-assisted review generated around forty defect reports, a task that would have been handled gradually over the following five years. Each of these reports requires a human committer to review, confirm, write a fix, test, and get it reviewed, a process that consumes a significant amount of time.

Notably, the SQL Property Graph Queries (SQL/PGQ) feature was removed due to unresolved issues, and its removal was forecasted to lead to post-release bug discoveries. Despite these cuts, PostgreSQL 19 is still expected to be a strong release, featuring REPACK CONCURRENTLY, which allows for reclaiming disk space without heavy table locking.

The delayed release affects PostgreSQL 20 as well, as it will likely include the reverted features from PostgreSQL 19. This situation highlights the emerging challenge faced by software projects worldwide, where AI reviewers accelerate the detection of bugs but also increase the work required for remediation.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in AI

Chinese stocks turn biggest losers as global AI trade stumbles

Underperformance of tech stocks poses a challenge for authorities keen to use capital markets to fuel firms’ growth

  • Chinese stocks plummet 40% in half year, worst-performing major tech gauge
  • MetaX Integrated Circuits and Moore Threads Technology lose half value
  • Outflows from Hong Kong trading link total 90 billion yuan in Q3

Comparing AI video models side by side: what I learned planning one prompt through Veo, Kling and Seedance

Disclosure: I work on Vidlark. All prices and specs below are copied from the public model pages as of October 2026. When people compare AI video models, they usually post three clips and a gut…

  • Veo generates 8-second clips in 720p or 1080p, with sound quality crucial for output
  • Kling 3.0 excels at capturing physical motion, recommends starting with low tracking shot
  • Seedance 2.0 is multi-shot, maintains consistency in people and props across generated clips

More from Saturday 17 October →