model-writtenrun 4942 - may be wrong, links go to the source messages. Commit-by-commit LLM analysis is in the Commits dialog.
- Peter Eisentraut proposed removing the long-deprecated postmaster executable alias because Meson and apparently MSVC did not install it, submitting startup-script cleanup and symlink/manual-page removal together in #1.
- Joe Conway supported removal but identified PGDG RPM service files that still invoked postmaster in #2; Devrim GÜNDÜZ valued the distinct parent-process name in #3, and Tom Lane warned about breaking startup scripts in #4.
- Andres Freund emphasized the lengthy deprecation and explained Meson's symlink portability complications in #5, while Robert Haas explained how the parent process remained identifiable and questioned indefinite deprecation in #6; Tom maintained that build-system convenience alone was insufficient justification in #8.
- Andres considered retaining the alias through platform-dependent symlink or copy behavior in #9, while Daniel Gustafsson argued against adding support for a supposedly deprecated feature in #10.
- Karl O. Pinc reviewed both patches, checked documentation references, successfully built the documentation, and identified an obsolete entity in doc/src/sgml/ref/allfiles.sgml in #11; he also flagged ambiguous runtime.sgml wording about invoking postmaster.
- Tom Lane opposed removing postmaster terminology for the controlling process in #12, Joe agreed in #13, and Karl clarified that the proposal removed the executable alias rather than the process terminology in #14.
- Karl reported successful installation and uninstallation tests and judged a broader directory rename unnecessary and potentially disruptive in #17.
- Peter argued that removing the alias would improve consistency across packaging and SELinux configurations in #18; Joe reaffirmed support in #19, while Devrim noted that Red Hat packages also invoked postmaster in #20.
- Peter sought resolution of the remaining packaging concern in #21, and Devrim withdrew any objection while indicating that RPMs might create their own compatibility symlink in #22.
- After Karl reiterated the missing SGML entity deletion in #23, Peter landed the first part of #1 with additional startup-script comment cleanup as 81266442fb and the second part with the entity deletion as 37e2673350, announcing the correction in #24.
- Karl revisited the ambiguous runtime.sgml wording in #25, prompting Peter to clarify executable references and nearby server terminology in 17e72ec45d; Peter confirmed that follow-up in the final message, #26, leaving both submitted patches and the documentation correction landed.