PLC writes request. Application writes acknowledgement. Clearing is the disputed bit. PLC author says acknowledgement means finished, application author says it means received.

Those meanings need separate names on your sheet. I would draw one successful inspection with arrows between the two programs, because the disagreement may become obvious without touching either program.

We had that exact language fight on a different machine. Receiving the job looked like completing it on the operator screen. Lovely quick cycle time, until somebody wanted an inspected part.

I've split received and result in the draft. Found another hole. Nobody has described what happens if the application restarts after accepting a request.

Do you already have job numbers? I'd use those in the drawing too, even a pretend number on paper. Otherwise restart gets discussed like everyone remembers the same job.

Just a request bit today. The inspection record gets a number later. So no shared identity at acceptance.

That makes recovery difficult. I'd ask the two authors to describe an interrupted inspection and an unknown result explicitly, with nobody automatically resending work just because a connection comes back.

Nathan, is your immediate task agreeing the interface or implementing it? A proposed identifier may require changes on both sides and deserves its own estimate.

Agreeing it. No budget for implementation approved yet. I can show the missing cases without pretending an extra field is free.

Paper exercise is enough for what I meant. My first quote for a small interface change grew once the records screen needed the same number. Still glad we asked before ordering.

Will your operators see received and finished as different states? Ours had one cheerful tick, which made the programming argument everybody else's problem.

Current screen has one completion lamp. Proposed screen has accepted, inspecting, result available and unknown. That last one has caused more discussion than the rest combined.

I would expect that discussion to include who investigates an unknown result. A label is useful, but somebody still needs time and access to reconcile the job.

Both reviewed the draft. Each signal has one writer now and normal clearing is agreed. Restart and timeout handling are still marked undecided. Approval stays partial.

I'd be interested in the restart decision when you reach it. That's the bit I never managed to make feel small in our own estimate.