diff options
| author | Craig Jennings <c@cjennings.net> | 2026-04-19 16:15:03 -0500 |
|---|---|---|
| committer | Craig Jennings <c@cjennings.net> | 2026-04-19 16:15:03 -0500 |
| commit | f926e6921311826b668e3e3a9fc2156b35cb1e77 (patch) | |
| tree | 872a72d62651003abb50a6e79786313d54fea953 /arch-decide/LICENSE | |
| parent | b11cfd66b185a253fecf10ad06080ae165f32a74 (diff) | |
| download | rulesets-f926e6921311826b668e3e3a9fc2156b35cb1e77.tar.gz rulesets-f926e6921311826b668e3e3a9fc2156b35cb1e77.zip | |
chore(add-tests): sharpen the pairwise-vs-exhaustive prompt
Previous wording — 'pairwise or continue with normal category coverage' —
muddled the choice. Category coverage (Normal/Boundary/Error) tests each
parameter individually and doesn't meaningfully address combinatorial
interaction questions.
Sharper framing: when a function has parameters that combine, the real
question is pairwise vs exhaustive (the M^N full factorial). Category
coverage is a separate axis — it handles the error-case and boundary-case
testing that parameter-value combinations don't.
New prompt cites specific counts:
'Function X has N parameters (M^N = Y exhaustive combinations).
Pairwise or exhaustive?'
Pairwise stays the pragmatic default; exhaustive only when the user names
a regulatory / safety-critical reason.
Diffstat (limited to 'arch-decide/LICENSE')
0 files changed, 0 insertions, 0 deletions