model-writtenrun 5943 - may be wrong, links go to the source messages. Commit-by-commit LLM analysis is in the Commits dialog.
- Peter Eisentraut proposed four RELKIND grouping macros to make repeated relation-kind checks clearer, accompanied by removal of redundant pg_dump inheritance checks, in #1.
- Alvaro Herrera recommended checking table AMs before other storage and questioned how partitioned-table AM inheritance would affect the macros in #2; Michael Paquier identified related work and suggested keeping the inheritance debate separate in #3.
- Peter discovered that RelationGetNumberOfBlocksInFork() incorrectly grouped partitioned indexes with storage-bearing objects and submitted a standalone correction in #4; Michael and Alvaro confirmed it in #5 and #6, and it landed unchanged as 0d906b2c0b.
- Peter revised the main series in #7, adopting table-AM-first handling, dropping RELKIND_HAS_INDEX_AM as conceptually unhelpful, and extending the refactoring to pgstattuple and relation-size estimation while retaining the interpretation that table AMs imply storage.
- After Peter rebased the series in #8, Michael supported the macro refactoring but argued that the explicit pg_dump checks were useful documentation and that heap_create() should preserve defensive checks against creating storage for unsupported relation kinds in #9.
- Peter proposed adding the missing materialized-view check instead of removing the pg_dump checks in #10; Alvaro supplied the explicit storage guard and OidIsValid correction and otherwise judged the refactoring positively in #11.
- Alvaro submitted an additional heap_create() simplification on top of the rebased series in #12, then withdrew it as nonworking in #13, so that delta was not incorporated.
- Michael endorsed the missing materialized-view check in #14, and Peter committed that discussion-derived change as a22d6a2cb6, declaring the original pg_dump removal patch obsolete in #15.
- Peter committed the macro-refactoring portion of #8 with the defensive storage guard and OidIsValid adjustment as 37b2764593, confirming completion in #16 while declining the suggested comment grammar change; the thread ended with the retained work landed and the alternative pg_dump removal and withdrawn storage simplification left out.