model-writtenrun 5865 - may be wrong, links go to the source messages. Commit-by-commit LLM analysis is in the Commits dialog.
- Jelte Fennema-Nio proposed treating ETIMEDOUT as a hard connection failure after observing that a low tcp_user_timeout could leave libpq failing to mark a connection CONNECTION_BAD #1.
- Tom Lane identified missing Windows error mappings and questioned the existing portability guard for ETIMEDOUT #2; Jelte added the Windows header mapping, but the revised attachment still lacked TranslateSocketError handling #3.
- Tom committed #3 as b484ddf4d2, adding the missing socket-error translation, removing the strerror portability guard, and making cosmetic adjustments; he announced that the change had landed on HEAD only #4, with the commit message citing Windows ABI concerns as a reason to defer backpatching.
- Jelte explained that the missing translation was present locally but omitted from his attachment, then requested a backport because the bug contributed to high query timeouts in Citus and supplied a variant conditional on TCP_USER_TIMEOUT to avoid changing the Windows ABI #5.
- Tom objected that the proposed conditional required c.h to include the header defining TCP_USER_TIMEOUT, which he considered undesirable #6.
- Nearly three months later, Jelte renewed the backport request with a variant that enables the change everywhere except Windows #7; the supplied thread ends with the development-branch fix committed and this backport proposal unresolved.