Our FR5 reporting application counts a retained completion again after reconnect. I know an uncertain attempt should not be resent, and nothing here resubmits work. The problem is the count: the accompanying register identifies a recipe, not the individual coupon check. Two rows can therefore look alike without being the same attempt.
The application has its previous saved results, but the PLC view currently exposed to it has only recipe and flag. The cell owner is checking whether a separate attempt history exists.
Then don't deduplicate by recipe. Mark the reconnection interval for reconciliation while the owner establishes what identifies individual outcomes; repeated observations and genuinely repeated checks need different treatment.
A completion count on our report once looked settled until the physical items were compared. Chen, does the current display tell its reader that the interval is unresolved, or does it still show the duplicated total as final?