SaraAbbott0026 · 20 Mar 2026, 04:45 UTC
We found ours was accepting any method name. Start by making unknown calls and wrong arguments fail against the documented interface for your installed SDK version.
16 pointsDeveloper replaced the catch-all behaviour with the documented methods we actually use. The misspelled call now fails. Response parsing is still barely tested.
15 pointsGive it representative documented success, failure and incomplete replies, then check what the application does with each. Returning a convenient true value bypasses the parser you are trying to exercise.
15 pointsWhat happens if you remove the result-read call entirely? Does opening the screen still count as a pass?
13 pointsYes. That test still passes without a read. I have added that as a known failing case for the developer.
21 pointsOur display retained a previous result after the next read failed. Sequence tests caught it; separate screenshots of success and failure did not.
1 pointsLiam's sequence is worth adding. Also start with no successful reading, so an old default cannot pretend to be the last known result.
12 pointsAnd distinguish a failed read from an actual inspection rejection. They can both be non-successful outcomes for the software, but the person at the bench needs different explanations.
2 pointsThe success-then-failure sequence exposed a stale accepted colour. Developer fixed it and added current-read validity. Missing and incomplete replies now show no usable result, not rejected bracket.
13 pointsDid the no-read test get fixed as well, Sara? Easy to lose that one once the colours behave.
13 pointsYes. It requires the expected read and validates the parsed result. Removing the call fails. Unknown method, wrong arguments, incomplete reply and initial no-result cases also pass their intended checks.
0 pointsHave the tutor read the failure wording. We thought unavailable was clear until somebody took it to mean the bracket was physically missing from the fixture.
7 pointsTutor distinguished read failed from rejected bracket in the offline sequence. Mock and parser test work is complete. Installed interface comparison remains for commissioning; these tests make no claim about physical handling.
10 pointsKeep the deliberately broken examples with the tests. They show why this substitute is useful much better than a page of green results alone.
2 pointsBy creating an account, you agree to our Terms and Conditions and community guidelines. Read our Privacy Policy for how your information is handled.