model-writtenrun 4271 - may be wrong, links go to the source messages. Commit-by-commit LLM analysis is in the Commits dialog.
- Michael Paquier proposed making unaccent regression tests portable to databases without UTF8 encoding by skipping their UTF8-dependent checks and adding alternate expected output, following the approach used elsewhere in the tree #1.
- Michael revised the patch to remove the now-unnecessary encoding and locale overrides from Makefile and meson.build, and said he intended to wait until development for version 17 opened #2.
- Peter Eisentraut endorsed making these checks consistent with the other test suites #3.
- In the thread’s final message, Michael thanked Peter for the review and explicitly left the encoding and locale settings in test_oat_hooks and test_extensions outside this change #4.
- The final patchset #2 subsequently landed unchanged in 44e73a498c, crediting Peter’s review and explaining that no backpatch was planned because the longstanding issue had attracted no complaints.