model-writtenrun 1048 - may be wrong, links go to the source messages. Commit-by-commit LLM analysis is in the Commits dialog.
- Chao Li proposed replacing raw exit(1) with proc_exit(1) in WalRcvWaitForStartPosition() so the walreceiver would execute normal cleanup callbacks #1.
- Andreas Karlsson supported the change because WalReceiverMain() already used proc_exit(1) for comparable shutdown states #2.
- Xuneng Zhou suggested exit status 0 for a requested shutdown #3, but Fujii Masao preferred retaining status 1 to match the SIGTERM-triggered ereport(FATAL) path and explained that the postmaster treats both statuses as non-crash exits #4.
- Fujii also identified an outdated termination comment #4; Chao first changed its wording to “terminate” #5, then adopted Fujii’s more precise “ereport(FATAL)” suggestion #6 and reflowed the comment in the final revision #7.
- Fujii committed the final revision #7 unchanged as ee550254a2, applying it only to master because no practical issues had been observed.
- Fujii announced the push #8, and Chao acknowledged it #9, closing the thread with the patch fully landed.