On 12/03/2017 02:15 PM, Adrian Bunk wrote: >> This looks like more >> an nondeterministic test than a hard failure, meaning that IMO, this bug >> should be downgraded to important. > > A flaky test that fails frequently is also an RC bug, imagine for > example a security update that FTBFS most of the time.
Right. It just depends on how frequently. If it's 1% of the time, I wouldn't make it RC. Anyway, let's not discuss this, and focus on fixing the problem. >> Could you get the reproducible machine to restart the build? I'd like to >> know if it's always failing on your machine, influenced by the fact you >> have more CPU cores than I, so it builds on 16 threads instead of 4... > > It shouldn't FTBFS with any number if CPU cores. Sure, I was just trying to understand what the issue was. >> Going forward, I'll probably blacklist this test so it doesn't fail >> again "sometimes". > > Yes, that sounds like a good solution. This will be done shortly. Thanks for your bug report. Cheers, Thomas Goirand (zigo)
