On Tue, Feb 04, 2020 at 04:14:14PM -0500, Kurt Mosiejczuk wrote: > On Tue, Feb 04, 2020 at 09:06:48PM +0100, Martin Reindl wrote: > > > Indeed, the py-nose TEST_DEPENDS was unnecessary, thanks for spotting. > > Surprisingly, with ports-gcc, test_memory_leak does not fail. I wonder > > how tests behave on sparc64, it builds OK there according to the > > packages dir. > > py-nose is needed. It *shouldn't* be, but a number of the tests require it. > (The tests could be rewritten to just use py-test). > > Once py-nose is in TEST_DEPENDS, all tests pass on sparc64.
Thanks for testing Kurt. Can you please explain where this comes from? RELEASE.rst states it is not needed anymore since 1.3.0 because the test suite now uses pytest. Also see: https://github.com/pydata/bottleneck/pull/222 In addition, I see no other reference in the code to py-nose, or when building with PORTS_PRIVSEP. But maybe I am overlooking something. -m
