Why does our FR5 dashboard look healthy when its request fails?

AnnaBell0677 · 16 May 2025, 02:39 UTC

Closed
AN
AnnaBell0677
The replacement laptop returns a container from our FR5 status call where the old environment returned a value, and our dashboard treats any nonempty answer as success; who should decide what gets displayed when the request contains an error?

14 replies

BR
BrunoBrooks0800
Replying to AnnaBell0677

Start with the documented response for the exact library version. Is there a success or error field the display code is currently skipping? My first question would be what the application actually received on one failed request.

9 points
AN
AnnaBell0677
Replying to BrunoBrooks0800

The wrapper documentation names a result flag and a data field; the programmer confirms the display currently checks only whether the returned object is nonempty.

18 points
BR
BrunoBrooks0800
Replying to AnnaBell0677

There it is. Have the programmer check that flag before using the data. A full envelope can contain bad news just as easily as good news.

16 points
CA
CalebBaker0493
Replying to AnnaBell0677

What's left on the screen after an error? Ours once kept a lovely green value from the last successful read. Very reassuring. Completely out of date.

8 points
BR
BrunoBrooks0800
Replying to CalebBaker0493

I'd want to see its age and that the new read failed. Otherwise someone can mistake a retained number for a current one, especially when nothing on the screen changes.

20 points
AN
AnnaBell0677
Replying to CalebBaker0493

It retains the old value with no age shown; I have added that to the fault description rather than treating the flag check as the whole repair.

13 points
BR
BrunoBrooks0800
Replying to AnnaBell0677

Good catch by Caleb. Are you testing the old environment too, or replacing it completely? Two return formats could need different handling before they reach the same display.

6 points
RA
RaviAdams0128
Replying to AnnaBell0677

Test a good result followed by an error, not just an error on startup; that's how the old value gets stranded on the screen.

10 points
AN
AnnaBell0677
Replying to BrunoBrooks0800

Both environments remain in use for now, and the developer is adding a small adapter for each documented format plus that good-then-error test.

23 points
AA
AaronBrown0871
Replying to AnnaBell0677

What happens if a response has neither the expected fields nor a recognisable error? I would want that to display as unavailable or unknown, since a future library change could otherwise restore the same misleading green state.

21 points
BR
BrunoBrooks0800
Replying to AaronBrown0871

And show a plain message for whoever uses it. Unknown response is enough on the main display; the developer can have the ugly details in the log.

22 points
AN
AnnaBell0677
Replying to AaronBrown0871

Unknown responses now show unavailable, and the stored last-good value is labelled with its age; the offline examples pass for both documented formats.

11 points
AN
AnnaBell0677
Replying to AnnaBell0677

The workshop check remains outstanding. At least we now have a failed response that fails visibly, an unexpectedly useful improvement for a status display.

19 points
BR
BrunoBrooks0800
Replying to AnnaBell0677

Let the workshop user try it before closing the repair. They should be able to tell an old reading from a new one without remembering this conversation.

4 points

Discussion closed

This discussion is closed to new replies after six months without activity. Last activity: .