mirror of
https://github.com/kuhyx/testsAndMisc.git
synced 2026-11-10 20:27:02 +01:00
Krzysztof kuhy Rudnicki
3ddc13f671
The largest file in phone_focus_mode, and the last one blocking deploy.sh:
splitting it grows deploy.sh's two hardcoded lists, so those lists are only
final now. All 55 functions moved verbatim.
Three while-read loops were audited before extraction. Two call pm in their
bodies and moved untouched. The third, iter_whitelist_packages, is fed by a
pipe from printf and reads nothing from stdin -- already safe, and now said
so in ctl_usage.sh's header so the distinction is not lost by a later editor
who has only read the warning in daemon_apps.sh.
136 assertions and 22 mutations, all killed. Three survived the first run
and every one was a defect in MY OWN tests, not in the subject:
- cmd_status carries a THIRD copy of the Haversine, and I had asserted
only that it printed "some number". A wrong Earth radius sailed through.
Now the value is checked against a seeded fix ~4.3 km from home.
- the two whitelist filters were each tested with input the OTHER filter
also rejects, so either could be deleted with nothing failing. Fixed
with a punctuation-free undotted token (only the dot check rejects it)
and a dotted token carrying an illegal character (only the charset check
does), making each independently observable.
That last investigation turned up dead code: the `#*)` case arm is provably
equivalent, because `set -- $line` makes a comment's first token "#", which
the dot check already rejects. Removed rather than covered by a test that
cannot fail.
Coverage is NOT 100% and the real numbers are in the evidence: 41.79% to
92.98% across the eight, up from 5-22%. Unlike the daemon split, most of
what remains is genuinely untested -- the per-enforcer status blocks that
print mount state, APK hashes and offload settings, each needing its own
device fixture. The mutations prove the assertions that exist are load-
bearing; they do not prove those branches work.
jscpd checked in a clean worktree at 1.57%: eight files each carrying a
near-identical start/stop/status trio is exactly what that gate is for, so
the tests drive the trios through tables instead of repeating them.
Two drift hazards recorded, not fixed: ctl_is_curfew_now duplicates
daemon_location.sh's is_curfew_now (pinned by an agreement test across every
boundary minute), and cmd_status's awk is the third Haversine.
Over-cap count 17 -> 16. deploy.sh is the last one here.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011w2Pj4Ek3ZNMQVYcFHgXDY
|
||
|---|---|---|
| .. | ||
| fixtures | ||
| lib | ||
| run_all | ||
| _schema_validation.py | ||
| _session_autopsy_mutations.py | ||
| check_agent_contract.sh | ||
| check_ai_evidence.sh | ||
| check_append_only_sessions.sh | ||
| check_file_length.sh | ||
| check_no_binaries.sh | ||
| check_no_dufs_conf.sh | ||
| check_no_inline_python.sh | ||
| check_no_secrets.sh | ||
| check_polling_antipatterns.sh | ||
| check_python_location.sh | ||
| check_requirements_sync.py | ||
| ci_mirror.sh | ||
| disk_cleanup_check.sh | ||
| extract_shell_functions.py | ||
| init_agent_artifacts.sh | ||
| install_hooks.sh | ||
| makepkg-manual-dlagent.sh | ||
| mutate_session_autopsy_hooks.py | ||
| mutate_shell.py | ||
| pytest_changed_packages.py | ||
| run_prettier_capped.sh | ||
| run_with_heavy_lock.sh | ||
| shell_coverage_report.py | ||
| shell_coverage.sh | ||
| trace_shell_split.sh | ||
| validate_contract.py | ||
| validate_evidence.py | ||
| verify_shell_split.sh | ||