model-writtenrun 8888 - may be wrong, links go to the source messages. Commit-by-commit LLM analysis is in the Commits dialog.
- Peter Eisentraut proposed a centralized make update-unicode workflow in #1, with Unicode and CLDR versions in src/Makefile.global.in, release-checklist instructions, and a generator replacing the embedded combining-character table.
- John Naylor tested the patch on macOS and reported missing library prerequisites and an unset Python interpreter in #2; he also recommended ignoring downloaded files, removing them with distclean, updating the README, and avoiding pgindent on generated output.
- Peter's revision in #3 added build prerequisites, corrected ignore and cleanup handling, and updated documentation, while retaining the short generated-file header and deferring pgindent changes because separate normalization work might resolve the formatting issue.
- John reviewed the revision and marked it ready in #4; Peter then committed it as f85a485f89, adding a Python fallback and a download dependency, and announced completion in #5.
- Tom Lane reported that the new header failed standalone compilation checks in #6; Peter explained its function-local inclusion in #7, and Tom proposed explicit exceptions in both header-checking scripts in #8.
- Peter implemented those exceptions in cc25464763 and confirmed completion in #9, resolving the reported failure without restructuring the header.
- By the final message, both changes had landed, and Peter reported the first Unicode data update using the new facility in #10, recommending annual updates around April to align with CLDR releases and PostgreSQL release preparation.