PostgreSQL 19开发期间大量功能被回退

PostgreSQL 19分支在测试阶段经历了异常多的功能撤销，共有8至10项重要功能被回退1。这些被取消的功能包括REPACK CONCURRENTLY、逻辑复制序列和pg_plan_advice等1。

导致功能回退的原因多方面。代码质量不足、复杂性过高是主要问题，而LLM辅助代码审查工具的引入进一步提升了质量审查标准，在测试中发现了大量边界情况问题1。此外，发生于4月至8月间的Vulnpocalypse安全事件分散了高级开发者的精力，对开发进度造成了冲击1。系统架构的复杂性也给测试带来了挑战1。

虽然多个功能被撤销，但PostgreSQL 19预期仍将是一个强大的版本1。开发社区认为许多被回退的功能将在后续版本中重新引入1。为应对日益增长的复杂性问题，开发者提议探索建立"实验性"功能状态机制1。

PostgreSQL's development branch for version 19 has undergone an exceptionally high number of feature reversions during its testing phase, with approximately 8 to 10 significant functionalities being withdrawn.1 The reversions stem from multiple interconnected factors that have impacted the development cycle's quality standards and resource allocation.

Code quality concerns have emerged as a primary driver behind the withdrawals.1 The implementation of AI-assisted code review processes has raised the effective quality threshold for what constitutes acceptable contributions, leading to the identification of numerous edge case issues that might have previously gone undetected.1 Simultaneously, the Vulnpocalypse security incident, which occurred between April and August, diverted the attention and efforts of senior developers away from feature development work.1 These combined pressures, along with the inherent complexity of PostgreSQL's architecture and the testing challenges it presents, have created an environment where maintaining code standards necessitated stepping back on multiple fronts.

Despite these reversions, the development community remains optimistic about PostgreSQL 19's prospects.1 The version is expected to remain a robust release, with many of the withdrawn features anticipated to be reintroduced in subsequent versions.1 Among the features that have survived the review process and remain part of the planned release are REPACK CONCURRENTLY, logical replication for sequences, and pg_plan_advice.1 In response to ongoing complexity challenges, developers have proposed exploring an "experimental" feature status mechanism as a potential framework for managing similarly intricate functionality in the future.1