Andrew Warshall wrote (Monday, November 05, 2012 4:19 PM): > On Mon, 5 Nov 2012 15:46:32 +0000 > "Voelker, Bernhard" <bernhard.voel...@siemens-enterprise.com> wrote: >> $ sudo env PATH="$PATH" make NON_ROOT_USERNAME=nobody check-TESTS >> PASS: [...] >> ../build-aux/test-driver: line 95: 5158 Aborted >> "$@" > $log_file 2>&1 FAIL: test-getlogin > > You've uncovered bug #11150, I believe. Is this in an xterm? The > getlogin test fails for me too there (at least, it's done so for a > while), and arguably rightly so. Basically, all that getlogin() does is > to look the TTY it's connected to up in /var/run/utmp. This should list > who's logged in at every terminal. Unfortunately, xterm doesn't write > to /var/run/utmp, so getlogin() fails in this case. Whether you think > this counts as a bug in getlogin() is up to you; anyway, it's different > (I assume) from the issue we were having. > > If you want to complain about the getlogin() issue, by the way, > that should go to the bug-gnulib list (at least, that's where it got > forwarded before).
Well, I don't mind. Actually, I find it more important to know whether gnulib tests should be run as root or not: are the gnulib tests secure enough to be run as root? I don't know, but I don't think a dependant project like coreutils should run them as root. Have a nice day, Berny
