model-writtenrun 6197 - may be wrong, links go to the source messages. Commit-by-commit LLM analysis is in the Commits dialog.
- Peter Eisentraut asked why verify_heapam rejected sequences, observing that removing the checks appeared to work #1.
- Mark Dilger supplied a preliminary WIP attachment and explained that portable corruption tests were difficult, while relying solely on existing heap tests might miss future changes to sequence storage #2.
- Peter Eisentraut suggested that checking an uncorrupted sequence could itself detect incompatible storage changes by unexpectedly reporting corruption #3.
- Mark Dilger adopted tests that exercise sequence creation, advancement, setting, and restarting, and submitted a patch enabling sequence checks in amcheck and adding sequences to pg_amcheck's table-pattern selection; he documented the behavior and asked whether treating sequences like tables without separate filtering switches was acceptable #4.
- Peter Eisentraut committed the submitted patch unchanged apart from rebase context as c3b011d991 and announced completion, leaving no outstanding work stated in the thread #5.