our fr5 dashboard finds an idle status inside a failed read

That's what happened. The old position now supplies a diagnostic value, and our local label table turns it into idle. Developer has the version and both successful and failed responses.

Does the report save that invented idle too?

Yes. Display and report both receive the wrapper's converted value. I've added the report row to the reproduction; it isn't only a misleading screen colour.

What will the operator see instead when there isn't a valid current reading?

Use an explicit unavailable result from the adapter and keep any previous reading separate with its observation time. Do not pass a made-up ordinary status downstream and expect each screen or export to recognise the trick.

Hana, would you remove the previous reading altogether? I'm not sure whether keeping it helps the operator or leaves another tempting number on the screen.

I'd ask the operator what they use it for, Dinesh, but I wouldn't leave it in the current-status position and hope a little timestamp explains the difference

Operator uses the last reading when reporting a stop. We've kept it in a separate previous-reading section, with current reading unavailable above. They could explain both on the revised test screen.

Has the parser fix passed, not just the wording?

Retained successful states parse correctly. Failed, incomplete and unexpected response shapes return unavailable. The same cases reach display and report without an ordinary status being invented. Installed workstation check remains.

Include first startup without any previous reading, and a good read followed by failure. Those are different screen histories even though both end without current data.

Did the workstation comparison settle it, Ben?

Yes. Maintainer checked the identified installed package with both startup cases and the retained response set. Current and previous status stay distinct, and the report records failed reads as unavailable. Operator handover accepted.

Thanks for keeping the previous reading useful without dressing it up as the current one; that is the bit I would want the cover operator shown too