[WaylandYang]

Every OWL import already writes a full ledger — classes created, updated, keys taken, inverse properties linked, attributes skipped and why, triples read, IRIs that did not project. The page printed the filename, the size, and who imported it when, and dropped the rest on the floor. After an import, "what actually came in" had no answer anywhere in the interface.

The history is now a table, one row per import, with the counts that are read at a glance: classes, relations, attributes, triples, when and by whom. Each cell reads created 12 · updated 3 · skipped 2 rather than a bare number, because "12 classes" is not the same statement as "12 new classes".

The rest is behind Details: the whole summary in two columns, plus the unprojected IRIs with their occurrence counts. Referring to an external vocabulary is normal and the projection is deliberately partial, so "how much did we not link, and to what" has to be sayable — but only someone chasing a problem reads it, and it does not deserve table width.

attributes_skipped is a reason → count map, not a number. The table sums it; the detail lists each reason with a translated label. It was previously typed as a number, which rendered [object Object] skipped.

🤖 Generated with Claude Code