Skip to content

Checking and correcting records

Capturing means appending. A stored record is never overwritten because it was stored — it is corrected when somebody corrects it, and that stays visible.

Everything that happens to a record goes through the same record tag:

Operation What happens
Append a new record is created (this is what the capture mask does)
Update a named record is corrected
Deactivate the record is retired
Reactivate it counts again

There is no delete. Captured records point at master-data rows; if a row disappears they point at nothing, and no evaluation over time could be trusted afterwards. Deactivating keeps the row readable and takes it out of the suggestion lists.

In the Record table — a widget on a dashboard. It shows the rows behind a record tag and offers, depending on what was granted: create a row, change one, retire one, bring one back. A Show inactive checkbox brings the retired rows into view.

The data view on the record type itself is read only. To change something you need the record table.

A widget reads by default. Whoever sets it up registers separately what it may do — create, update, deactivate — and which role may use it. The three are separate because they are separate risks: correcting a typo in a defect name is not the same act as retiring that defect.

What you see is always what the server granted. With nothing registered the table is read-only for everybody, a tenant administrator included.

  • The whole row is written, not a single field. An optional field left empty is therefore really cleared.
  • The values are checked exactly as at capture time — datatype, required fields, closed value lists. A correction is made at a desk with the current state on screen; there is no leniency here for late-arriving entries.
  • Hidden fields are not part of the form and keep their values.

A correction and every (de)activation write an audit entry: who, when, on which record, which fields. The values are not in the audit — it is a trail, not a second copy of the data.

Appending is not audited: a captured record carries its own provenance, and a running capture stream would drown the trail the audit exists to protect.

The capture time stays the capture time. A correction is stamped with the time of the correction — it records when something was put right, and does not rewrite when it was taken down.