model-writtenrun 2155 - may be wrong, links go to the source messages. Commit-by-commit LLM analysis is in the Commits dialog.
- Hayato Kuroda proposed allowing pg_recvlogical to create failover-enabled slots so administrators could test slot synchronization, supplying --failover/-O, documentation, and a regression test in #1.
- Michael Banck suggested omitting a short option in #2, and Hayato adopted that in #3; Masahiko Sawada then requested clearer, alphabetically ordered documentation in #4, which Hayato incorporated with revised help text in #5.
- Masahiko committed #5 as cf2655a902 with light help-text and formatting changes, announcing the push in #6.
- Zhijie Hou found the surviving --falover documentation typo and supplied a one-line follow-up in #7, which Masahiko committed unchanged as a13d49014d and acknowledged in #8.
- In June, Peter Eisentraut reopened the interface discussion because --failover sounded like an action, proposing --enable-failover in #9; Masahiko raised the corresponding --two-phase name in #10, and Amit Kapila favored treating both consistently while discussing existing API naming in #11.
- Hayato accepted either naming choice but proposed alternative two-phase documentation wording in #12; Peter pointed to pg_createsubscriber's --enable-two-phase precedent in #13, and Masahiko supplied the two-option rename in #14, crediting Peter as author and leaving Hayato's wording suggestion unadopted.
- Peter reviewed the rename in #15 and requested retaining --two-phase as a deprecated alias to avoid breaking existing users.
- Peter committed #14 as 6d12d5a433, adding the compatibility alias, documenting it, retaining a test of the old spelling, and adjusting error strings; the new --failover spelling received no compatibility alias because it had not appeared in a released version.
- Peter announced the final commit in #16, and Masahiko thanked him for taking over in #17, leaving all submitted changes landed with no further patch pending.