ChenAllen0268 · 5 May 2025, 05:17 UTC
Compare an actual success and failure with the installed SDK documentation; don't guess which item in the returned object is the status
11 pointsOur developer found the whole response was being treated as true because it was non-empty. Thank you, that gave us a useful starting check.
6 pointsThere's your misleading green.
21 pointsThe replacement needs more than a test of one error. Try a valid response, a reported failure, a missing value and an unexpected shape. Failed interpretation should not turn into a cheerful default. And show when the last good value is old.
6 pointsYes, old data needs to look old.
18 pointsThe draft now checks the documented success field before using the data. A failed read shows current status unavailable, with any older reading labelled stale.
11 pointsCan the operator tell those states apart without relying on colour?
13 pointsOn another dashboard, text labels fixed the colour problem but the page still refreshed the old value's timestamp; make sure that timestamp belongs to the observation, not the latest attempt to fetch it.
8 pointsUseful catch. Our timestamp was page refresh time. The developer has changed it to the last successful observation, and the states have explicit text labels.
14 pointsThe recorded success and failure cases now display correctly, including malformed replies and stale data. The updated read-only dashboard is in use. We kept the captured examples with the SDK version.
8 pointsMuch better than green by accident.
19 pointsThis discussion is closed to new replies after six months without activity. Last activity: .