Thank you Ole! Indeed taking an attempt at having them actually fixed is preferable, otherwise those bugs tend to accumulate. Cheers
On January 26, 2019 8:38:16 AM EST, Ole Streicher <oleb...@debian.org> wrote: >Ole Streicher <oleb...@debian.org> writes: >> since some time, scikit-learn is in danger to be removed from testing >> (#907806), but fails to migrate now due to several test failures on >> non-intel architectures (#919918). I opened an upstream issue >> >> https://github.com/scikit-learn/scikit-learn/issues/13036 >> >> which suggests that some of the test failures are simple to fix, some >> others however seem to be serious. There are some packages depending >on >> scikit-learn, that's why I would like to push this forward a bit. So, >I >> would like no to just disable all failing tests to ensure the package >> migrates. We can re-enable these tests once they are fixed upstream. >> >> Any objections? Specific ideas how to fix one or the other failure >> without disabling it? > >Since upstream reports that some of them are fixed with 0.20.2, I would >at first attempt to upgrade to that version tonight. > >Cheers > >Ole -- Yaroslav O. Halchenko (mobile version) Center for Open Neuroscience http://centerforopenneuroscience.org Dartmouth College, NH, USA