I tried to compile coreutils-8.14-1 from abs, with standard Arch CFLAGS. It fails the "make check" in one test, test-parse-datetime. I found a recent reference<http://lists.gnu.org/archive/html/coreutils/2011-10/msg00099.html>to a patch needed to avoid false failures in this test caused by incorrect handling of DST. (I'm not on DST right now). Shouldn't the coreutils in abs be passing this test during the build? Is it safe to ignore the failure? [BTW, coreutils fails to compile with "-march=native" on an i5-2405s (native translates to corei7-avx plus a bunch of flags for sse4 etc.) but succeeds (except for that one test) by adding "-mno-avx" to CFLAGS. "Fails to compile" is an understatement. While compiling coreutils I briefly saw what looked like an old ROM debugger screen and then my display came back but all of my file systems were read-only. I think folks might want to add the "-mno-avx" flag to "-march=native" on most Sandy Bridge builds unless you know that the package can take advantage of the avx instruction set. Firefox has been known to fail at run time when compiled for "native" on Sandy Bridge.] -- sitquietly
