| Summary: | sys-apps/sed-4.2 - tests utf8-[1-4] fail ("XPASS") | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Jeroen Roovers (RETIRED) <jer> |
| Component: | Current packages | Assignee: | Gentoo's Team for Core System packages <base-system> |
| Status: | CONFIRMED --- | ||
| Severity: | normal | CC: | esqualante, fauli, gentoo, jakub.januszkiewicz, rhill |
| Priority: | High | Keywords: | TESTFAILURE |
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux | ||
| URL: | http://lists.gnu.org/archive/html/bug-gnu-utils/2010-04/msg00012.html | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: | sys-apps:sed-4.2:20090630-175303.log [hppa,fail] | ||
|
Description
Jeroen Roovers (RETIRED)
2009-06-30 19:00:33 UTC
Created attachment 196197 [details]
sys-apps:sed-4.2:20090630-175303.log [hppa,fail]
Same here on x86, alpha is ok. It is triggered by non-English locales. jer has nl, I have de while Tobias succeeded with en. LC_ALL=C makes the test suite run flawlessly. (In reply to comment #3) > It is triggered by non-English locales. jer has nl, I have de while Tobias > succeeded with en. LC_ALL=C makes the test suite run flawlessly. No, it's triggered by any UTF8 locale, regardless of language. (In reply to comment #4) > (In reply to comment #3) > > It is triggered by non-English locales. jer has nl, I have de while Tobias > > succeeded with en. LC_ALL=C makes the test suite run flawlessly. > > No, it's triggered by any UTF8 locale, regardless of language. > Actually, it's triggered by a *lack* of the ru_RU.UTF-8 locale, but presence of any other UTF-8 locale. This is because of an incorrect check in the configure script for UTF-8 support, which basically checks to see if Cyrillic case folding functions properly. It does this by trying to switch to the ru_RU.UTF-8 locale, then confirming that towupper() works on a lower/upper case pair of Cyrillic characters. The problem is that, if you don't have ru_RU.UTF-8 available, the test switching explicitly to the "C" locale, then immediately checks to see if the locale name ends in .UTF8. Since this will always be false, the configure test *always* fails if you don't have ru_RU.UTF-8 available. So it sets up the test suite to expect all the utf8-* tests to fail. The test suite test itself, apparently, does the correct thing and switches to whatever available UTF-8 locale you have defined by LC_ALL, and the tests pass. A quick workaround is to add ru_RU.UTF-8 to /etc/locale.gen and re-run locale-gen. A better fix is for someone who understands the setlocale() stuff better than I do, to fix the test in configure.ac, just after "AC_MSG_CHECKING([whether UTF-8 case folding tests should pass])". |