Am 04.02.20 um 14:28 schrieb Charlene Wendling: > On Tue, 4 Feb 2020 09:11:07 +0100 > Martin Reindl wrote: > >> Hello ports@ >> >> attached diff updates math/py-bottleneck to 1.3.1: >> >> - take MAINTAINER >> - BROKEN-powerpc: there have been some changes in upstream, please >> retest on powerpc and sparc64, for now I've added COMPILER but we >> might be able to do without! >> - tests all pass on python3, with python2 one test out of 190 fails >> (amd64 tested) >> >> OK? >> >> -m > > I tried on powerpc and: > > - it builds and passes all tests, see the logs [0] > - it fails with the same error BROKEN-powerpc message when base-gcc > is used, so that COMPILER line is needed. I've moved it per > Makefile.template recommendations, and added a reason for it > - py-nose was not picked up because TEST_DEPENDS is redefined later, > i fixed that > > Below is a diff with these changes made. As far as powerpc goes, i'm > fine with it. The diff looks good to me but i know nothing about python > porting subtleties. > > Charlène. > > > [0] https://bin.charlenew.xyz/py-bottleneck.tgz
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. -m
