On Sun, 10 Jun 2018 at 21:57 Paul Gevers <elb...@debian.org> wrote: > Hi > > On 10-06-18 21:34, Tristan Seligmann wrote: > > -vectors exists purely for the benefit of the tests, nothing outside the > > tests uses the vectors. If you are not running the tests, having a > > mismatched version of -vectors will have no effect. > > Thanks for clarifying. Just for my understanding, did you ever > considered to add these vectors (as a second tar ball in the source 3 > format) to the source package of python-cryptography? >
I think this might actually be the ideal solution, all else being equal, but a lot of the tooling in use (dh_python, pybuild, git-dpm, etc.) unfortunately does not handle this scenario well, so I haven't attempted it.