Dataset for ICSE 2026 Paper “Test Flimsiness: Characterizing Flakiness Induced by Mutation to the Code Under Test”

This archive contains the full dataset that is too large to include in the main replication package. Filename conventions make use of the subject number (subject_n), the mutant number (mutant_n), and the run number (run_n). The mutant number -1 indicates no mutation. Negative run numbers indicate coverage runs. Contents are as follows:volume/coverage: Compressed Coverage.py reports for every coverage run. Filename convention is as follows: "{subject_n}_{mutant_n}_{run_n}.coverage.gz".volume/diffs: Mutant diffs. Filename convention is as follows: "{subject_n}_{mutant_n}.diff".volume/metrics: Compressed runtime metrics about each test case for every test suite run. Not used in the study. Filename convention is as follows: "{subject_n}_{mutant_n}_{run_n}.pkl.gz".volume/mutants: Cosmic Ray arguments to produce each mutant. Filename convention is as follows: "{subject_n}.csv".volume/reports: Compressed JUnit XML reports for every test suite run. Filename convention is as follows: "{subject_n}_{mutant_n}_{run_n}.xml.gz".volume/sources: Source code of each test case. Filename convention is as follows: "{subject_n}.pkl".