model-writtenrun 4331 - may be wrong, links go to the source messages. Commit-by-commit LLM analysis is in the Commits dialog.
- Bruce Momjian opened the thread while writing PostgreSQL 16 release notes, identifying confusing membership directions in createuser's --role and --member options and proposing clearer SQL and utility documentation in #1.
- Daniel Gustafsson supported the documentation patch and proposed removing parentheses around the group explanation in #2; Bruce initially defended them in #3, while Robert Haas highlighted the underlying ROLE versus IN ROLE ambiguity in #4.
- Peter Eisentraut suggested --member-of to clarify direction in #5, and Michael Paquier supported exploring that name in #6 and separately accepted the documentation changes in #7.
- Bruce incorporated Daniel's parentheses change in #8, while Nathan Bossart clarified that --member-of should replace --role because it makes the new role a member of an existing role in #9.
- Bruce committed the revised documentation from #8 unchanged as ce3673dade and announced it in #10.
- Peter proposed distinct --with-members and --member-of names in #12; Nathan agreed that --role must remain compatible in #13 and supplied #14, using --with-member, --member-of, and --with-admin with regression tests.
- Bruce raised the beta1 timing question in #15, and Tom Lane supported landing before beta1 while requesting clearer prose, help text, and direct reuse of the 'g' option handler in #16; Nathan incorporated those changes in #17.
- Tom endorsed the revision with another documentation wording correction and accepted leaving --role undocumented in #18, and Michael reviewed the revision and compatibility test in #20.
- Nathan committed #17 as 2dcd1578c4, with Tom's final wording correction and small test-ordering changes, and announced completion in #21.
- Peter objected that leaving --role undocumented conflicted with the release's treatment of deprecated options in #22, prompting Nathan's follow-up patch #23 to restore it to documentation and help with a deprecation label.
- Michael accepted the follow-up and advised waiting for the beta1 stamp in #24; Nathan committed #23 unchanged as 381d19b3ea and confirmed completion in the final message #25, leaving all submitted work landed.