two attempts sharing a very unhelpful job label

Restart is clear, sequence number isn't. Python starts numbering again at one, while the exported PLC trace only contains the job label and acknowledgement state

Ask controls whether the original capture contains more fields than the export. You might recover an attempt reference that was left out. Keep the original order within each log meanwhile, even though the clocks do not line up.

Julia's point matters for the report too. A state sampled at capture start needs a different description from a recorded transition. Otherwise the next person may read it as an event you actually saw.

Already on in the first sample. I had drawn an arrow from it to the retry, which now looks a bit ambitious. Removing that arrow

Keep the observation, though. It tells the reader what was present when capture began, even if its origin remains unknown. Put the missing earlier trace beside that limitation.

We once spent ages aligning clocks only to find a held bit in every row. Is there an older capture on the engineering laptop before anyone goes looking for clever timing tools?

Did controls find any earlier capture or extra identifier, Anika? I'm curious whether you managed to connect either attempt, or whether this remains a logging change for the next investigation.

No earlier capture turned up. Controls is adding an attempt reference that survives application restarts; the old acknowledgement stays unassigned. Thanks Julia, that arrow would have become a very confident paragraph