The view from here: I've now got an LSB-conforming platform to run the SI on top of, which means I *should* be able to eliminate that as a source of errors (hmmm).
I've located several glibc patches from the glibc cvs which improve standards conformance, eliminating the ftw, readv/writev (assuming the kernel *also* has the fixes for those), and other failures. There's a patch (two lines changed) for fileutils which eliminated the T.misc chown failure. Turns out this was a known bug in the released fileutils 4.1 tarball. Sean, I'll forward these two patches to you separately. The nice tests continue to fail which I don't understand yet, it's as if the getpriority call didn't work: the four failing tests all come back with a zero for the priority instead of what it should have been set to. There are some syslog tests that fail when run under the SI and not when run on the native system, using the same syslogd binary. I presume this is a library issue, but I haven't found it. getlogin continues to fail apparently due to running on top of an already logged in terminal. At some point I'll try running something like sshd to enter the SI, but no time right now. A setlocale test is Unresolved (really a fail, but Unres because the failure is in a place the test doesn't know what to do with it), I'm waiting for some expert advice on this one (I believe I do know how to fix it, it would be removing one character from a glibc file). There is, I believe, one other fail and one unresolved in the 32-bit LSB-si as I'm currently running it. I'm running an abbreviated test now, removing the two signal tests (accounting for 160 minutes of the total test time of 6:42), so I might have a new result by the time of the call. Mats -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with subject of "unsubscribe". Trouble? Email [EMAIL PROTECTED]
