# Test fixtures are regenerated by scripts (test_golden_regression.py
# --update-golden, test_clustering_e2e.py). Python's text-mode writes turn "\n"
# into "\r\n" on Windows, so without this every regeneration rewrites every line
# of every fixture and produces a whitespace-only diff of ~900k lines.
# Pin them to LF in both the repository and the working tree.
*.csv text eol=lf
*.json text eol=lf
