AI-Assisted Work Advances 10-Year Drupal Core Issue on Field Widget Ordering

The issue stems from Drupal’s default use of tabledrag for all multi-value field widgets, even when item reordering is unnecessary. According to Tanton, this behaviour complicates theming, adds JavaScript overhead, and creates accessibility and responsive layout challenges. For years, the only workaround was a large, fragile patch that required constant maintenance across core updates.

The proposed solution introduces an orderable boolean setting, allowing individual widgets to opt out of drag-and-drop behaviour. To support this cleanly, the work adds a shared configuration schema so widgets can inherit the setting without duplicating definitions. The approach is designed to reduce long-term maintenance while providing site builders with greater control over form behaviour.

Tanton explains that AI tools were used as a development aid rather than a decision-maker. They assisted with drafting a clearer issue summary, exploring architectural alternatives, expanding test coverage, and iterating on Twig templates and CSS. In several cases, screenshots were fed into the AI workflow to speed up UI adjustments that would otherwise require multiple manual passes.

The project relied on structured workflows, including plan-driven development and a dedicated context directory to preserve continuity between sessions. Despite these efficiencies, Tanton repeatedly stresses that expert oversight remained essential. He notes instances where AI-generated code relied on weak assumptions or required careful correction through manual review and testing.

The issue has not yet been committed to Drupal core, but its status has improved significantly. The remaining work is largely focused on theming refinements and documentation polish ahead of final review. Tanton presents the effort as a practical example of where AI can accelerate routine but complex contribution tasks, while cautioning against treating automated output as authoritative.