Our FR5 inspection application adds one coupon when it reconnects to an already-high completion signal; I have kept the attempt identities this time, so I can show that the restored completion belongs to an item already counted, but the reconnect handler still treats it as new work.
Show the maintainer the same completion before and after reconnect. A counter correction alone could leave the next connection ready to make the same mistake.
Use that replay to test repeated reconnects. Once is too kind to it. Also include a retained completion whose identity cannot be matched; fixing this known coupon mustn't teach the app to quietly discard every awkward result.
Yes. Otherwise the operator gets a wonderfully calm screen while maintenance inherits a mystery later. Use wording that says what is unresolved, not another success colour.
Both display and export, Jonas; Gabriel, the replay now includes repeated reconnects and an unmatched retained result, with the latter kept visible outside the accepted total.
Does that state survive restarting the application as well? Reconnecting within one session might be using memory that the next startup won't have. I'd put both in the review before calling the reporting bug fixed.
The revised replay passes both restart cases, Gabriel; Caleb, reconciliation still needs an agreed owner and procedure, so I have not handed this version over as finished.