Experimental Feature: Root Cause Clusters in Fossil Test

[dreamer-coding]

We’re excited to introduce an experimental root cause clustering feature in Fossil Test! This new addition leverages TIM/TI failure analysis to automatically group similar test failures and suggest potential causes. Each failure is categorized into common problem areas such as Memory, I/O, Logic, Assumption, Timeout, or Unknown, making it easier to identify patterns across large test suites. The clustering logic tracks failure occurrences, samples representative failures, and provides hints for possible resolutions, helping developers focus their debugging efforts efficiently.

The feature works by hashing individual failures and attempting to group them into clusters based on similarity. Each cluster maintains its own list of failures along with metadata like file, function, and line number. A helper routine guesses the root cause of each failure based on error messages and filenames. Once clustered, Fossil Test generates a concise report showing the number of occurrences per cluster, sample failure details, and actionable suggestions. This experimental approach is inspired by feedback from fossil_test_summary_feedback and aims to improve test triage without adding manual overhead.

We encourage users to experiment with this feature and share feedback. Because it’s still experimental, there are limits on the number of clusters and failures per cluster, and the categorization heuristics may evolve over time. By analyzing failure patterns in this structured way, teams can uncover hidden root causes, reduce repeated debugging cycles, and ultimately improve test reliability. Try it out, check the clustered reports, and let us know how it changes your workflow!