model-writtenrun 528 - may be wrong, links go to the source messages. Commit-by-commit LLM analysis is in the Commits dialog.
- Karina Litskevich proposed using int64 for pg_stat_statements entry counts and persisted counts in PostgreSQL 19 and later, plus correcting pgver to uint32; she also described an alternative for older releases that preserved their file format in #1.
- Daniel Gustafsson supported the change after beta2 and suggested leaving older released branches untouched because a practical problem was unlikely in #2, which Karina accepted in #3.
- Karina resent the unchanged main patch alone after cfbot tried to apply the alternative patches together in #4; Daniel confirmed that CI was green in #5, and Tom Lane explained how the .nocfbot suffix could exclude alternatives in #6.
- Daniel supplied a revised two-patch series in #7, separating the unrelated pgver correction, adding explicit int64 casts for pgss_max, and moving the int64 loop variable into its loop declaration.
- Karina accepted Daniel's changes in #9, leaving the series agreed but not yet committed at the thread's last message.
- Daniel subsequently committed the first patch from #7 unchanged as 3374850360 and backpatched it to PostgreSQL 19, then committed the second patch unchanged as 1b87dd4303; both changes retained Karina's authorship, and the older-release alternative was left out.