mirror of
https://github.com/kuhyx/testsAndMisc.git
synced 2026-11-11 19:14:02 +01:00
Krzysztof kuhy Rudnicki
b91be3099d
The first allowlist removal of the campaign this session. unity_handler.sh measures 32/32 = 100.00% through its directory's runner, and check_shell_coverage.sh passes it with the exemption DELETED -- the real gate, not just the measurement. HARNESS BUG found getting there: bash caches executable locations in a hash table, so `command -v foo` keeps succeeding after foo is removed from the stub dir. Every "tool is missing" test was therefore running as a "tool is present" test, asserting the opposite of its own name while looking plausible. Reproduced minimally: after rm, STILL FOUND; after `hash -r`, gone. A _t_unstub helper now removes the stub and clears the table. The two suites committed an hour ago had the same defect and are corrected here -- test_transcribe_pkgmgr.sh goes from 15 to 18 passing assertions, so three of its checks had been vacuous. This is exactly the failure the campaign exists to catch, found in its own test code. `basename` joins the restricted PATH because the SUBJECT calls it; the suite avoids needing `tail` via a parameter expansion instead, since every binary in that PATH is one more thing a stub's shebang can trip over. transcribe_deps.sh (92.31%) and transcribe_pkgmgr.sh (98.00%) re-measured unchanged and stay on the allowlist. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_016sxxsDHWgdRhmdmzNMA6XM
52 lines
4.0 KiB
JSON
52 lines
4.0 KiB
JSON
{
|
|
"intent": "Cover unity_handler.sh to a measured 100% and take it OFF the shell-coverage allowlist -- the first entry removed this session, 105 -> 104. Fix the bash hash-cache bug found doing it, which had silently weakened the two suites committed just before.",
|
|
"scope": [
|
|
".../testsAndMisc-bash/lib/tests/test_unity_handler.sh: new, 18 assertions across all 4 functions",
|
|
".../tests/transcribe_harness.sh: new _t_unstub helper; basename added to the restricted PATH",
|
|
".../tests/test_transcribe_deps.sh and test_transcribe_pkgmgr.sh: switched to _t_unstub",
|
|
"meta/shell-coverage-allowlist.txt: unity_handler.sh removed",
|
|
"Non-goal: no change to any subject; no latent bug surfaced in unity_handler.sh"
|
|
],
|
|
"changes": [
|
|
"HARNESS BUG, found by an assertion that would not go green: bash caches executable locations in a hash table, so `command -v foo` keeps SUCCEEDING after foo is deleted from the stub dir. Every 'tool is missing' test therefore ran as a 'tool is present' test -- asserting the opposite of its own name while looking plausible. Reproduced minimally: after `rm`, STILL FOUND; after `hash -r`, gone.",
|
|
"The fix is a _t_unstub helper that removes the stub and clears the table, applied to all three suites in the directory. The two committed an hour earlier were affected and are corrected here; test_transcribe_pkgmgr.sh goes 15 -> 18 passing assertions as a result, so three of its checks had been vacuous.",
|
|
"`basename` was added to the restricted PATH because the SUBJECT calls it, unlike bash/env which the stubs' shebangs need. The suite avoids needing `tail` by taking the last line with a parameter expansion instead: every binary in that PATH is another thing a shebang can trip over.",
|
|
"The generated .desktop file is asserted on its CONTENT (Exec and MimeType lines), and xdg-mime registration on the BASENAME it is given -- a full path there would silently fail to register."
|
|
],
|
|
"verification": [
|
|
{
|
|
"command": "shell_coverage_jail.sh --subject .../tests/run_all.sh --measure unity_handler.sh --min 100 --fail-on-case-error",
|
|
"result": "pass",
|
|
"evidence": "unity_handler.sh: 32/32 lines = 100.00%, exit 0 at --min 100. Measured through run_all.sh, the subject is_covered() builds."
|
|
},
|
|
{
|
|
"command": "bash meta/scripts/check_shell_coverage.sh .../unity_handler.sh (AFTER removing it from the allowlist)",
|
|
"result": "pass",
|
|
"evidence": "'every checked shell library meets the coverage bar', exit 0. The lib passes the REAL gate with no exemption -- properly cleared, not merely measured."
|
|
},
|
|
{
|
|
"command": ".../lib/tests/run_all.sh",
|
|
"result": "pass",
|
|
"evidence": "15 + 18 + 18 = 51 assertions, 0 failures, 'All transcribe lib test files passed.'"
|
|
},
|
|
{
|
|
"command": "minimal repro of the hash cache: stub a command, run it, rm it, then command -v",
|
|
"result": "pass",
|
|
"evidence": "'STILL FOUND (hash cache)' before `hash -r`, 'gone after hash -r'. This is what makes _t_unstub necessary rather than cosmetic."
|
|
},
|
|
{
|
|
"command": "re-measure the other two libs after the harness change",
|
|
"result": "pass",
|
|
"evidence": "transcribe_deps.sh 48/52 = 92.31% and transcribe_pkgmgr.sh 49/50 = 98.00%, both unchanged. Both stay on the allowlist."
|
|
}
|
|
],
|
|
"risks": [
|
|
"Any other suite in the repo that removes a stub with a bare `rm` has the same latent defect: its 'missing tool' branch is not being tested. The features/lib suites use a jail and mostly add rather than remove stubs, but this is worth a sweep before the gate is armed.",
|
|
"The restricted PATH must list every binary the subjects use. A missing one surfaces as 'command not found' mid-suite rather than as a clear assertion failure."
|
|
],
|
|
"rollback": [
|
|
"git revert the commit; unity_handler.sh returns to the allowlist and the three suites return to bare `rm` unstubbing.",
|
|
"After rollback, confirm the allowlist is back to 105 entries so the gate's view is unchanged."
|
|
]
|
|
}
|