I reopen the monitor and the accepted total grows by one, although no new bracket has been checked. The PLC still holds the last completion and our reconnect routine treats seeing it as a new event.
I want the saved result matched to its job before the total changes. Clearing the retained value would make the screen quieter, but also lose what we're trying to reconcile.
Check what identity is retained with the result and what the monitor has already saved. Does that identify an inspection attempt, or only a bracket that could be checked again?
Only bracket identity in the current retained record. The monitor saves result rows, but reconnect doesn't compare them before counting. Retried brackets make a simple duplicate-label filter wrong too.
Then keep ambiguous matches out of the accepted total and agree unique attempt identity for the revised exchange. The historical cases still need their own disposition review.
What does completed mean in that record? Receipt of the request, finished inspection and accepted bracket are three different things the screen could accidentally merge.
Finished inspection, with a separate acceptance result. We do check that result before counting; the problem is counting the same accepted attempt again. Thanks Luis, worth stating precisely.
Include restart between receiving a result and saving its handled status in the tests. Otherwise the new attempt identity can still arrive twice at the counter.
Maintainer has added that case. Proposed version keeps handled attempts persistently and leaves uncertain history for review. No automatic resubmission is part of this monitor.
Our monitor passed reconnect but failed closing and reopening the whole application. Give that its own test with both a known handled result and an unknown result, not just the same happy example twice.
Tell the shift what the review state means as well, Nina, because an accurate incomplete total is useful only if people don't keep reopening the program until it says the number they expect
Known handled result survives application restart without another count. Unknown result goes to review. Failed-save testing is unfinished, so the new version is not handed over. Jamie, the draft screen explains that review does not mean run the inspection again.