Incident Report: ArrayIndexOutOfBoundsException When Synchronizing JTable Sort Order to JList in Java File Browser

Summary

A developer attempted to synchronize sorting between a JTable and a JList by capturing the sorted view indices via TableRowSorter during a sort event. They encountered an ArrayIndexOutOfBoundsException when iterating over row indices, preventing the synchronization of the file browser components.

Root Cause

Why This Happens in Real Systems

• Filtering overhead: Many systems combine sorting with dynamic filtering, shrinking the visible row count.
• Event timing confusion: The SORTED event fires during structural updates, making assumptions about row counts risky.
• State mismatch: Failing to distinguish between model-level and view-level counts causes synchronization errors.
• Overloaded components: Integrating multiple synchronized UI components (JTable/JList) complicates state sharing.

Real-World Impact

• UI inconsistency: Failure to synchronize sorting broke user expectations of unified file listings.
• Application crashes: Uncaught exceptions terminated critical UI workflows.
• Data corruption risk: Out-of-bounds accesses risked memory-adjacent data corruption.
• Debugging difficulty: Timing-dependent errors produced inconsistent stack traces.

Example or Code

Incorrect Implementation:

The `RowSorterListener` handler incorrectly used the **model row count** (`sorter.getModelRowCount()`) to determine the iteration bounds for view indices. When the sorted view contained fewer rows than the model (e.g., due to filtering), attempting to access view indices beyond the actual view row count caused the index-out-of-bounds exception.

java
sorter.addRowSorterListener(e -> {
if (e.getType() == RowSorterEvent.Type.SORTED) {
// Problem: Uses MODEL row count for VIEW index iteration
int[] modelIndex = new int[sorter.getModelRowCount()];
for (int i = 0; i < sorter.getModelRowCount(); i++) {
// Fails if view has fewer rows than model
modelIndex[i] = sorter.convertRowIndexToModel(i);
}
}
});

How Senior Engineers Fix It

• Use view bounds: Replace getModelRowCount() with sorter.getViewRowCount() in iteration.
• Leverage existing APIs: Clone the sorted view indices via sorter.getViewToModel().
• Decouple logic: Move sorting state to the underlying model, not UI components.
• Add fail-safes:
  • Validate indices with sorter.getViewRowCount()
  • Wrap in try-catch for edge-case recovery
• Refactor synchronization: Implement a shared RowSorter or custom SortedListModel.

Corrected Code:
java
sorter.addRowSorterListener(e -> {
if (e.getType() == RowSorterEvent.Type.SORTED) {
List sortedModelIndices = new ArrayList<>();
for (int viewIndex = 0; viewIndex < sorter.getViewRowCount(); viewIndex++) {
sortedModelIndices.add(sorter.convertRowIndexToModel(viewIndex));
}
// Apply sortedModelIndices to JList here
}
});

Why Juniors Miss It

• Model-view confusion: Misunderstanding rigidity of model data vs. volatility of view data.
• API assumptions: Assuming getModelRowCount() returns view-safe values.
• Incomplete testing: Testing only with unfiltered data misses filtering edge cases.
• Event handling traps: Overestimating state stability within event callbacks.
• Duplicate effort syndrome: Trying to manually replicate sorting logic already managed by TableRowSorter.