Select the Available Items

The left-join report now gives each item one of two visible shapes:

That missing joined ID is the evidence we need. Open report.sql and add this line before ORDER BY:

WHERE loans.id IS NULL

Press Run. The report should contain the folding table, projector screen, and repair kit. The drill and projector have active matches, so their loan IDs are not missing and those items do not pass the new filter.

The folding table is the useful boundary case. It has loan history, but its old row does not match the active-only condition inside ON. The left join therefore keeps the item with missing loan columns. WHERE loans.id IS NULL selects it as available alongside items that have never been borrowed.

Do not replace this filter with loans.returned_on IS NULL after the join. An active matched loan already has a NULL return value, and an unmatched item also has NULL in every joined loan column. That test would keep both borrowed and available items. Testing loans.id IS NULL distinguishes them: a matched loan has an ID, while a missing match does not. Keep the active-loan condition inside ON; the WHERE clause asks whether any such match was found.

The result remains ordered by items.asset_tag:

IT-200 | Folding table | NULL
IT-400 | Projector screen | NULL
IT-500 | Repair kit | NULL

Submit shuffles insertion order and uses active, returned-only, and never-loaned items. It expects only the latter two groups. The SQL values are not tested with Python truthiness, and = NULL cannot substitute for IS NULL.

You can now answer availability without storing a second availability flag. The current answer comes from the presence or absence of an active relationship, which cannot drift away from the loan history it summarizes.