On Wed, 4 May 2016, Yaroslav Halchenko wrote: > severity 790925 important > severity 814795 important > thanks > > On Wed, 04 May 2016, Santiago Vila wrote: > > The package was removed from testing, so there is no "past" (unless we > > want to consider stable as the "past"). > > > In other words: Would the maintainer consider making #790925 and > > #814795 just "important" for pandas to propagate to testing on the > > architectures it currently builds? > > interesting take on things - I have never allowed myself such "free > willing" ;) > > FTBFS bugs usually are indeed 'serious' (thus RC-critical) by > default. In the case of pandas I can indeed make a case (upstream > doesn't care about non x86 archs atm) that it must not be unusable for > the rest of the ecosystem thus lowering it to important. So let's do it > and see how it goes - I don't think it would be sufficient alone for > migration to testing which does have memories of previous builds in > there.
Hi. In case it does not work as expected, there is also another thing that you could do, and it is your prerrogative as package maintainer, which is to modify "Architecture: any" in debian/control and put only the "actually supported" architectures. This would tell autobuilders not in the list that they should not try to build the package at all, and would also limit the "all architectures in sync" rule for unstable to testing migration to the architectures in the list. This would be done only as a last resort, and only if you lose completely all hope that it will work some day, because in principle, it is better that autobuilders try the build and fail, than they do not even try.
