model-writtenrun 7673 - may be wrong, links go to the source messages. Commit-by-commit LLM analysis is in the Commits dialog.
- David G. Johnston opened the thread with a documentation patch prompted by confusion about parameters in SET ROLE, aiming to clarify where PL/pgSQL variable substitution works and remove obsolete guidance #1.
- Pavel Stehule requested examples of utility commands and SQL identifiers #2, which David incorporated while reorganizing the explanation around utility commands followed by INSERT/UPDATE/DELETE #3.
- Pavel argued that execution-plan support, rather than whether a command returns results, explains parameterization #4; David preferred to leave the fuller explanation in the linked variable-substitution section #5.
- David Steele reviewed the wording and questioned references to PL/SQL #7, but Tom Lane explained that those references could intentionally describe Oracle's language #9, and Steele acknowledged that context #11.
- Tom took over revision of the patch, rejecting the utility-first explanation as confusing and introducing an explicit distinction between optimizable commands and utility commands requiring dynamic EXECUTE #12.
- Pavel questioned whether readers would understand optimizable-command terminology #13; Tom explained that DML/DDL categories mishandle cases such as CREATE TABLE AS SELECT #14, while David requested clearer exceptions and an explanation of identifiers versus substitutable values #16.
- Tom then reorganized the documentation into a general SQL-command section, separating substitution from output handling and explicitly including EXPLAIN and CREATE TABLE AS SELECT among commands supporting parameters #17.
- Tom committed that revision with additional explanations, terminology cleanup, and correction of the default ambiguous-name behavior in c783e656d4, crediting David Johnston as co-author and Pavel Stehule and David Steele as reviewers; his final message reported the completed landing and invited further comments #18.