From 91f498fdec14888d58190021fc125ec468d3047b Mon Sep 17 00:00:00 2001 From: Thomas Munro Date: Mon, 13 May 2024 07:55:20 +1200 Subject: [PATCH] Skip citext_utf8 test on Windows. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Back-patch of commit cff4e5a3 to 15 and 16, per request from Oleg Tselebrovskiy. Original commit message: On other Windows build farm animals it is already skipped because they don't use UTF-8 encoding. On "hamerkop", UTF-8 is used, and then the test fails. It is not clear to me (a non-Windows person looking only at buildfarm evidence) whether Windows is less sophisticated than other OSes and doesn't know how to downcase Turkish İ with the standard Unicode database, or if it is more sophisticated than other systems and uses locale-specific behavior like ICU does. Whichever the reason, the result is the same: we need to skip the test on Windows, just as we already do for ICU, at least until a Windows-savvy developer comes up with a better idea. The technique for detecting the OS is borrowed from collate.windows.win1252.sql. This was anticipated by commit c2e8bd27, but the problem only surfaced when Windows build farm animals started using Meson. Reviewed-by: Tom Lane Discussion: https://postgr.es/m/CA%2BhUKGJ1LeC3aE2qQYTK95rFVON3ZVoTQpTKJqxkHdtEyawH4A%40mail.gmail.com --- contrib/citext/expected/citext_utf8.out | 3 +++ contrib/citext/expected/citext_utf8_1.out | 3 +++ contrib/citext/sql/citext_utf8.sql | 3 +++ 3 files changed, 9 insertions(+) diff --git a/contrib/citext/expected/citext_utf8.out b/contrib/citext/expected/citext_utf8.out index 6630e09a4de..65bb70a79f9 100644 --- a/contrib/citext/expected/citext_utf8.out +++ b/contrib/citext/expected/citext_utf8.out @@ -6,8 +6,11 @@ * Turkish dotted I is not correct for many ICU locales. citext always * uses the default collation, so it's not easy to restrict the test * to the "tr-TR-x-icu" collation where it will succeed. + * + * Also disable for Windows. It fails similarly, at least in some locales. */ SELECT getdatabaseencoding() <> 'UTF8' OR + version() ~ '(Visual C\+\+|mingw32|windows)' OR (SELECT (datlocprovider = 'c' AND datctype = 'C') OR datlocprovider = 'i' FROM pg_database WHERE datname=current_database()) diff --git a/contrib/citext/expected/citext_utf8_1.out b/contrib/citext/expected/citext_utf8_1.out index 3caa7a00d42..525ad173914 100644 --- a/contrib/citext/expected/citext_utf8_1.out +++ b/contrib/citext/expected/citext_utf8_1.out @@ -6,8 +6,11 @@ * Turkish dotted I is not correct for many ICU locales. citext always * uses the default collation, so it's not easy to restrict the test * to the "tr-TR-x-icu" collation where it will succeed. + * + * Also disable for Windows. It fails similarly, at least in some locales. */ SELECT getdatabaseencoding() <> 'UTF8' OR + version() ~ '(Visual C\+\+|mingw32|windows)' OR (SELECT (datlocprovider = 'c' AND datctype = 'C') OR datlocprovider = 'i' FROM pg_database WHERE datname=current_database()) diff --git a/contrib/citext/sql/citext_utf8.sql b/contrib/citext/sql/citext_utf8.sql index 1f51df134bf..8c2ceca9558 100644 --- a/contrib/citext/sql/citext_utf8.sql +++ b/contrib/citext/sql/citext_utf8.sql @@ -6,9 +6,12 @@ * Turkish dotted I is not correct for many ICU locales. citext always * uses the default collation, so it's not easy to restrict the test * to the "tr-TR-x-icu" collation where it will succeed. + * + * Also disable for Windows. It fails similarly, at least in some locales. */ SELECT getdatabaseencoding() <> 'UTF8' OR + version() ~ '(Visual C\+\+|mingw32|windows)' OR (SELECT (datlocprovider = 'c' AND datctype = 'C') OR datlocprovider = 'i' FROM pg_database WHERE datname=current_database())