On 4/5/19 10:09 AM, Andreas Tille wrote: > On Fri, Apr 05, 2019 at 10:05:09AM +0200, Olivier Sallou wrote: >>>>> Frequently, on conda package update, it checks and propose to update >>>>> conda. If it is installed via debian package, it should not self update >>>>> from conda web site. >>>> Sure, conda packaged for Debian should run self updates. >>> s/should/should NOT/ >> I had understood ;-) > My correction was just for people without mind reading capabilities. ;-) > >> I looked at failing tests. >> >> Some *could* be skiped I think as related to issue to capture >> stdout/stderr , might be linked to build with pbuild/gbp , ... >> >> >> Some other tests of install/activate/deactivate fail however due to >> other setup stuff (don't find where/why for the moment). >> >> It tries to exec conda from /usr/bin/conda instead of *build >> prefix*/bin/conda >> >> And logs show that CONDA_PREFIX is set to >> >> * /usr > May be CONDA_PREFIX can be set for the build process in d/rules? > >> * /tmp/54d6 75cc (example) <= what looks strange here is there is a >> space in temp dir.... > +1 > >> It may explain that it tries to use /tmp prefix , fail for not finding >> it, then use fallback prefix /usr (and fail too because not installed on >> system, which is fine as it is not yet installed). >> >> >> before skipping those, I'd better understand why and if only related to >> tests condition > I fully agree. It might be perfectly possible that I skiped to much > of these. I'd be really happy about proof reading of my patches.
It tries to switch from different environments and fails at default one which is /usr (no write access) si I think tests calling self.basic_posix should be disabled in test_activate.py > > Thanks a lot for looking into this > > Andreas. > -- Olivier Sallou Univ Rennes, Inria, CNRS, IRISA Irisa, Campus de Beaulieu F-35042 RENNES - FRANCE Tel: 02.99.84.71.95 gpg key id: 4096R/326D8438 (keyring.debian.org) Key fingerprint = 5FB4 6F83 D3B9 5204 6335 D26D 78DC 68DB 326D 8438
