PostgreSQL 19 delayed to November as AI code reviews uncover more defects
Core developers have postponed the release of PostgreSQL 19 until November after LLM-assisted reviews identified a high volume of issues that require additional time to resolve.
Key points
- PostgreSQL 19 release is delayed from September/October to November 19.
- Between eight and ten significant features have been removed from this release cycle.
- Peter Eisentraut attributes the delay to improved LLM code review tools finding edge cases.
The PostgreSQL Global Development Group has officially delayed the general availability of PostgreSQL 19. Originally scheduled for late September or early October, the database version is now expected to launch around November 19. This shift comes after core developers decided to revert several planned features to address quality concerns.
Developers have dropped between eight and ten significant features from the upcoming release. These features will now be postponed to the next major version, PostgreSQL 20. The decision was made to ensure the stability of the core database system. The delay reflects a strategic choice to prioritize code quality over the original timeline for feature inclusion.
Impact of AI Code Reviews
Peter Eisentraut, vice president at EDB and a leading PostgreSQL committer, explained the reasons behind the reverts. He stated that some reverted features contained significant architectural defects. These issues would have been difficult to fix quickly during the beta testing phase. Eisentraut noted that the quality bar for code has effectively risen during the development cycle.
Eisentraut attributed part of the delay to the use of large language model (LLM) code reviews. These AI tools uncovered a long tail of relatively harmless issues involving various edge cases. In previous development cycles, developers might not have noticed these issues until years after launch. The AI tools found them much faster, creating a large list of defect reports.
According to Eisentraut, dealing with these reports creates a fixed overhead. Even if each issue requires only a small fix, such as three lines of code, the administrative burden adds up. The development team ran out of time to address the forty-plus defect reports generated by the AI reviews. This situation forced the team to cut features to meet the original timeline, which they chose not to do.
Development Tooling Evolution
The effectiveness of LLM-assisted code review has improved significantly in recent months. When the major updates for PostgreSQL 19 were initially written, these tools were less useful. However, as the launch date approached, AI-supported reviews became better understood and more effective. This improvement in tooling raised the standard for what constitutes acceptable code.
Eisentraut suggested that this situation should sort itself out for future development cycles. He noted that unless the capabilities of LLM-assisted code review continue to improve at the current rate, the process will stabilize. The current friction is a result of the tools catching up to the development workflow. The team is adapting to this new reality of higher detection rates for edge cases.
What should you do?
Users planning to upgrade to PostgreSQL 19 should adjust their schedules to account for the November 19 release date. Organizations relying on the removed features should evaluate alternative solutions or wait for PostgreSQL 20. Database administrators should monitor official PostgreSQL channels for further updates on the release timeline.
Background: Large language models
Large language models are a specific type of neural network trained on text. Generative AI is the broader umbrella term for any system that creates new content, including images and code. Not all generative AI uses large language models, but most modern text generators do.
What to do and how to stay safe
- Review your upgrade timeline to ensure it aligns with the new November 19 release date for PostgreSQL 19.
- Identify any planned implementations that depend on the eight to ten removed features and plan for their postponement.
- Monitor official PostgreSQL development blogs and announcements for updates on the final feature set and stability reports.
- Evaluate current database stability and consider if any edge-case issues identified by AI tools might affect your existing PostgreSQL versions.
Step-by-step guide: Prompt Injection Warning Signs: How to Spot AI Manipulation
General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
When is PostgreSQL 19 now expected to be released?
PostgreSQL 19 is now expected to be generally available around November 19, a delay from the original late September or early October target.
How many features were removed from PostgreSQL 19?
Developers have dropped between eight and ten significant features from PostgreSQL 19, postponing them to the PostgreSQL 20 release.
