One training laptop still uses it, Robin, so we need two known combinations during the changeover; the new package reply starts with a request status that happens to resemble a valid old mode value

Keep separate parsing functions if that makes the contracts obvious, then return the same application-level result to the display; a failed request needs an unavailable reading, not a guessed controller mode.

What does the screen actually do when the reply is empty, Rebecca? The numeric coincidence is awkward, but I would also want to see the less convincing failures.

It retains the previous mode without showing its age, Anika; the wrapper treats an empty reply differently from a reply carrying an error, so we have two misleading failure paths

Two bugs for the price of one upgrade.

Keep the last value labelled old, or clear it; don't present it as fresh.

I prefer retaining the last value with its time for a diagnostic screen, provided unavailable is unmistakable; ask the people using this screen which information they need during a failed read.

Ravi, I'd want them to try it with a stopped update, not just choose the nicer mock-up; a small old-value note can disappear beside a large confident mode label.

Maintenance chose unavailable as the main reading and a smaller last-observed value with its time; they found the original mock-up too easy to read as a current mode, which is exactly Anika's concern

What happens if the configured package and loaded package disagree?

The revised adapter refuses that combination instead of trying the other parser; tests cover both known formats, request errors, missing mode data, empty responses and the package mismatch

Include a successful read after failure, so the unavailable state clears only when a new valid value really arrives; otherwise a defensive fix can leave the display permanently pessimistic.

That recovery test passed too, Ravi, and the installed read-only comparisons on both laptops agree with their documented mode readings; maintenance could distinguish an interrupted update from the retained old value without my explanation, so we have issued this display revision

Is the old laptop's retirement on the handover list? Two supported formats need two future checks.