RaviBrown0911 · 24 Dec 2025, 10:07 UTC
What does the complete failed reply look like, with its type and SDK version? Keep a successful reply too. I'd want those beside the documented return description before deciding which part means status.
2 pointsDoes the current adapter check the request outcome before reading the status field, or does it treat any nonempty response as usable?
21 pointsAny nonempty reply, Harish. Daniel, maintainer has the full success and failure examples with the environment version; both containers pass that check
8 pointsThen the happy screen is entirely believable. The wrapper asked whether something arrived, not whether it contained a valid status. Stop that failed example reaching the mode display while the mapping is reviewed.
-7 pointsOur old panel kept the last good value after an error and forgot to say it was old. Will yours distinguish unavailable now from the previous successful read?
15 pointsFiona, agreed on withholding the failed reading. I'd make the test check the visible output as well as the adapter return, or another layer can quietly put the old plausible value back.
5 pointsUse the documented outcome and field types for that version, with unsupported or malformed responses rejected; a field position alone is not a stable interpretation.
7 pointsJack, it currently doesn't show age. Proposed display separates unavailable current status from the last good reading; maintainer is adding time and connection identity to that retained value
9 pointsThat would have saved us a pointless cable hunt. We were studying a perfectly steady number that hadn't changed because nobody was updating it.
18 pointsHave they tested starting with a failed read? There isn't a last good value then. And change connections in the test, so one controller's old reading can't dress up as another one's history.
19 pointsInclude the full malformed-response cases in the support handover. A future SDK change should fail visibly against those tests, instead of reviving the same guessing game with another container.
20 pointsInitial failure and changed connection both tested now. No invented first value; old readings stay linked to their source. Version mapping passes offline examples, installed comparison still pending
15 pointsWho owns that installed comparison and the adapter's supported-version record?
10 pointsSoftware maintainer with our cell technician. They've accepted the work, not completed it. Thanks Jack for the age question; our screen really did hide that
7 pointsPlease post what the installed comparison finds. Useful to know whether the captured examples covered the real responses, rather than only whether the revised tests were green.
19 pointsThis discussion is closed to new replies after six months without activity. Last activity: .