model-writtenrun 842 - may be wrong, links go to the source messages. Commit-by-commit LLM analysis is in the Commits dialog.
- Diego Frias reported that Unicode normalization incorrectly treated U+11A7 as a Hangul trailing consonant and supplied a one-line boundary fix, asking where a custom regression test should go #1.
- Michael Paquier confirmed that normalization swallowed U+11A7, identified the SQL regression suite as the appropriate test location, and posted Diego’s fix together with broader Hangul composition and decomposition tests #2.
- Diego approved the expanded patch and said he would try to contribute the tests to the official Unicode test suite #3.
- Michael committed the file changes from #2 unchanged as f2ff15e4c3, revising the commit message to distinguish Diego’s code contribution from his own tests.
- Michael closed the thread by confirming that the fix had been applied and backpatched through PostgreSQL 14, leaving no PostgreSQL changes pending #4.