testsAndMisc/docs/superpowers/evidence/transcribe-venv-off-allowlist-2026-08-22.json
Krzysztof kuhy Rudnicki 154b416617 Clear transcribe_venv.sh: 100%, off the allowlist, 104 -> 103
A 29-assertion suite covering all six functions. transcribe_venv.sh measures
53/53 = 100.00% through the directory runner, and check_shell_coverage.sh
passes it with the exemption deleted.

Assertions are on recorded ARGV because which index a pip install points at
IS the behaviour: the OpenNMT cu12 wheel index on the NVIDIA path, the
CPU-only torch index for diarization (forced to avoid mismatched CUDA
builds). "pip was called" cannot see either.

generate_test_audio's four-deep fallback chain is covered by failing each
generator in turn. Every stub materialises the wav, because the chain
advances on `[[ ! -s $tmpwav ]]` -- a record-only stub would silently
exercise the branch after the one under test.

One assertion was wrong and got corrected rather than forced: I claimed a
wholly-failing pip still returns 0, but the final `python "$PY_HELPERS"
deps-installed` carries no `|| log`, so it does not. The suite now asserts
what the subject actually guarantees -- all four warnings emitted, no earlier
failure aborting the run.

SC2016 fixed at the source with a quoted heredoc, not suppressed.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016sxxsDHWgdRhmdmzNMA6XM
2026-08-22 13:24:28 +02:00

45 lines
3.1 KiB
JSON

{
"intent": "Cover transcribe_venv.sh to a measured 100% and take it off the allowlist -- the second removal this session, 104 -> 103.",
"scope": [
".../testsAndMisc-bash/lib/tests/test_transcribe_venv.sh: new, 29 assertions across all 6 functions",
"meta/shell-coverage-allowlist.txt: transcribe_venv.sh removed",
"Non-goal: no change to the subject; no latent bug surfaced"
],
"changes": [
"Assertions are on RECORDED ARGV because which index a pip install points at IS the behaviour here: the OpenNMT cu12 wheel index on the NVIDIA path, and the CPU-only torch index for diarization (forced to avoid mismatched CUDA builds). A 'pip was called' check cannot see either.",
"generate_test_audio's four-deep fallback chain (espeak-ng -> espeak -> Python stdlib -> ffmpeg) is covered by failing each generator in turn. Each stub MATERIALISES the wav, because the chain advances on `[[ ! -s $tmpwav ]]` -- a record-only stub would silently exercise the next branch instead of the one under test.",
"The four `|| log \"Warning: ...\"` pip-failure paths are covered with a python stub that always exits 1. An initial assertion that the function still returns 0 was WRONG and was corrected rather than forced: the final `python \"$PY_HELPERS\" deps-installed` has no `|| log`, so a wholly-failing python does make it return non-zero. The suite now asserts what the subject actually guarantees -- all four warnings are emitted and no earlier failure aborts the run.",
"SC2016 was fixed at the source by switching the failing-pip stub to a quoted heredoc, not suppressed."
],
"verification": [
{
"command": "shell_coverage_jail.sh --subject .../tests/run_all.sh --measure transcribe_venv.sh --min 100 --fail-on-case-error",
"result": "pass",
"evidence": "transcribe_venv.sh: 53/53 lines = 100.00%, exit 0 at --min 100."
},
{
"command": "bash meta/scripts/check_shell_coverage.sh .../transcribe_venv.sh (AFTER removing the allowlist entry)",
"result": "pass",
"evidence": "'every checked shell library meets the coverage bar'. The real gate accepts it with no exemption."
},
{
"command": ".../lib/tests/run_all.sh",
"result": "pass",
"evidence": "15 + 18 + 29 + 18 = 80 assertions, 0 failures across all four suites in the directory."
},
{
"command": "shellcheck on the new suite",
"result": "pass",
"evidence": "clean; no suppressions added."
}
],
"risks": [
"The suite stubs python/python3 heavily. If the subject later stops routing work through $PY_HELPERS the stubs would keep passing while testing nothing -- the assertions are on argv, so a changed argv fails loudly, but a REMOVED call would not.",
"generate_test_audio's chain depends on each stub writing a non-empty file. A future stub that forgets to would move the test to a different branch than its name claims, the same failure mode this suite exists to avoid."
],
"rollback": [
"git revert the commit; transcribe_venv.sh returns to the allowlist and loses its suite.",
"After rollback, confirm the allowlist is back to 104 entries."
]
}