> > it would be easier to enforce to have all buildds configured equally so > > the package does not fail on a random buildd. > > I *guess* that's not trivial as I think it depends on the network their in.
OTOH it's not ideal to have machines that are supposed to be similar, being different enough to cause build failure (due to their configuration, not due to resource constraint). > m2crypto needs another upload anyways. Without a proper IPv6 solution > for the tests, can the failing network tests please be disabled in the > next upload? given the 29 failures are all part of test_ssl.py (and make up more than half of the tests in it) i'm going to skip that file entirely. i would argue that testing SSL is important for this package, so this solution is very suboptimal. -- Sandro "morph" Tosi My website: http://sandrotosi.me/ Me at Debian: http://wiki.debian.org/SandroTosi Twitter: https://twitter.com/sandrotosi
