On Mon, Mar 6, 2023 at 11:20 AM Jim Jones <jim.jo...@uni-muenster.de> wrote: > On 05.03.23 22:00, Thomas Munro wrote: > > could be something to do with > > our environment, since .cirrus.yml sets LANG=C in the 32 bit test run > > -- maybe try that locally?
> Also using LANGUAGE=C the result is the same for me - all tests pass > just fine. I couldn't reproduce that locally either, but I just tested on CI with your patch applied saw the failure, and then removed "PYTHONCOERCECLOCALE=0 LANG=C" and it's all green: https://github.com/macdice/postgres/commit/91999f5d13ac2df6f7237a301ed6cf73f2bb5b6d Without looking too closely, my first guess would have been that this just isn't going to work without UTF-8 database encoding, so you might need to skip the test (see for example src/test/regress/expected/unicode_1.out). It's annoying that "xml" already has 3 expected variants... hmm. BTW shouldn't it be failing in a more explicit way somewhere sooner if the database encoding is not UTF-8, rather than getting confused?