On 2 July 2016 at 09:05, Karen Etheridge <p...@froods.org> wrote: > > Hopefully there is enough information in these reports to diagnose the root > issue; if not, no doubt the collective wisdom exists on this list. thank > you!
+1 These seriously broken smoker setups degrade the overall quality of the CPAN Testers network very quickly, mostly because its non-trivial to isolate these spurious failures from legitimate failures that have to be fixed. If these failures are indicative of some broken CPAN tooling, I'm all for somebody working out what that is and fixing it. But in the meantime, these are just noise :/ Ideally the CPAN Testers system should be smart enough to categorise failures like this differently so we can handle them more effectively, but that is harder than fixing the problem at its root. For instance, maybe we could find some simple rules that marked a failure as "suspicious" if it can be determined that items in META.* were not satisfied. Or at very least, we should have some sort of system to mark smokers known to be "broken" automatically as "suspicious" until the smoker can be deemed "unbroken". After all, what is actually stopping somebody from writing tooling that abuses the CPAN reporters infrastructure to create entirely bogus failure reports, with the content being some religious proselyting? -- Kent KENTNL - https://metacpan.org/author/KENTNL
