Hello, previously I've reported on this mailing list about test failures. Now we've managed to localize them. So may be it's possible to fix some of them upstream?
1. smidump-sming test failure [1]. As Juergen told [2] it's Ok. Why don't you disable this test then? We already do this with the following command:
sed -i '/^[[:space:]]*smidump-sming.test \$/d' test/Makefile
2. smidump-orig-smiv2 failure[1]. This test failure caused by different locale settings and export LC_ALL=C fixes the problem [3]. May be it's good idea to add this into test suit itself?
3. smidump-cm fails. Initially it was reproduced on hppa, but later we found that it caused by compiler optimizations. So if you build with -Os or -O3 this test fails [4]. Now we forced building of libsmi only with -O2.
Will you accept patches for 1. and 2? Of course they are trivial... The test failure 3. is strange. Seems that code relies on some float pointing operations but code itself is not very reliable although I have not investigated it further and don't know where that code resides. What do you think?
[1] http://mail.ibr.cs.tu-bs.de/pipermail/libsmi/2008-March/001005.html [2] http://mail.ibr.cs.tu-bs.de/pipermail/libsmi/2008-March/001014.html [3] http://mail.ibr.cs.tu-bs.de/pipermail/libsmi/2008-March/001025.html [4] https://bugs.gentoo.org/show_bug.cgi?id=218541