AI code review just delayed PostgreSQL 19. Finding bugs is now cheap and fixing them isn’t.
PostgreSQL 19 was supposed to ship in October. The release team now says general availability is November 19, with RC 2 on November 12 (postgresql.org’s RC 1 announcement has the dates). The reason is the interesting part. What happened According to The Register’s report, developers reverted or pulled between 8 and 10 significant features and pushed them to PostgreSQL 20. Some of them, like…
PostgreSQL 19, originally slated for October release, now has a delayed general availability set for November 19, with the release candidate 2 (RC 2) scheduled for November 12. The primary reason behind this delay, according to The Register, is that developers reverted or pulled between 8 and 10 significant features and pushed them to PostgreSQL 20.
Some of these features, such as SQL/PGQ graph queries, faced architectural issues that proved difficult to resolve during the beta phase. Additionally, AI-assisted code review unearthed a substantial list of defects, with The Register citing roughly 40 defect reports, mostly concerning small edge cases that would otherwise have gone unnoticed or undetected.
While Peter Eisentraut, a PostgreSQL committer at EDB, clarified that "vibe coding" was not the cause of the reverts, as AI tools were not available during the initial development of most features, the AI did play a crucial role in identifying these defects. However, it's important to note that AI was not the sole factor; review capacity and the late-cycle review timing also contributed to the delay.
The underlying lesson here is that while AI has made defect detection more cost-effective, the actual fixing of these bugs remains a labor-intensive process. This means that the bottleneck has shifted: finding bugs has become cheaper, but fixing them is still expensive. For individuals maintaining a codebase, the practical takeaways from this story are clear: running AI reviews before merging code is essential, as late findings are far more costly.
Early defect detection is cheap and beneficial. It's also advisable to keep changes small, as a small diff leads to a smaller fix. Lastly, every fix must be thoroughly verified with tests, even when AI suggests a solution. The author of this report has been adopting these principles by favoring small, verified commits and developing Vigilante, a free game that rewards GitHub-verified commits as a collective settlement.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.