Repository navigation
pytest broken again - the return of #6424 #7409
Description
Activity
Thing is, at one point today it did work. I'm not sure what changed though. Damn gremlins!!!
conda 4.5.1
The following NEW packages will be INSTALLED: ca-certificates: 2018.03.07-0 certifi: 2018.4.16-py36_0 cython: 0.28.2-py36h1de35cc_0 intel-openmp: 2018.0.0-8 jinja2: 2.10-py36hd36f9c5_0 libcxx: 4.0.1-h579ed51_0 libcxxabi: 4.0.1-hebd6815_0 libedit: 3.1-hb4e282d_0 libffi: 3.2.1-h475c297_4 libgfortran: 3.0.1-h93005f0_2 markupsafe: 1.0-py36h3a1e703_1 mkl: 2018.0.2-1 ncurses: 6.0-hd04f020_2 numpy: 1.11.3-py36h8a80b8c_4 openssl: 1.0.2o-h26aff7b_0 pip: 9.0.3-py36_0 python: 3.6.5-hc167b69_1 readline: 7.0-hc1231fa_4 setuptools: 39.0.1-py36_0 sqlite: 3.23.1-hf1716c9_0 tk: 8.6.7-h35a86e2_3 wheel: 0.31.0-py36_0 xz: 5.2.3-h727817e_4 zlib: 1.2.11-hf3cbc9b_2I've been trying to reproduce but I suspect this might be a quirk in your local environment. Any other hints you can provide for reproducing would be helpful.
Here's what I get for the installed package list when running
conda build:astropy: 3.0.2-py36_1 local attrs: 18.1.0-py36_0 ca-certificates: 2018.03.07-0 certifi: 2018.4.16-py36_0 intel-openmp: 2018.0.0-8 libcxx: 4.0.1-h579ed51_0 libcxxabi: 4.0.1-hebd6815_0 libedit: 3.1.20170329-hb402a30_2 libffi: 3.2.1-h475c297_4 libgfortran: 3.0.1-h93005f0_2 mkl: 2018.0.2-1 mkl_fft: 1.0.1-py36h917ab60_0 mkl_random: 1.0.1-py36h78cc56f_0 more-itertools: 4.1.0-py36_0 ncurses: 6.1-h0a44026_0 numpy: 1.14.2-py36ha9ae307_1 openssl: 1.0.2o-h26aff7b_0 pip: 10.0.1-py36_0 pluggy: 0.6.0-py36hb1d0581_0 psutil: 5.4.5-py36h1de35cc_0 py: 1.5.3-py36_0 pytest: 3.5.1-py36_0 pytest-arraydiff: 0.2-py36_0 pytest-astropy: 0.3.0-py36_0 pytest-doctestplus: 0.1.3-py36_0 pytest-openfiles: 0.3.0-py36_0 pytest-remotedata: 0.2.1-py36_0 python: 3.6.5-hc167b69_1 readline: 7.0-hc1231fa_4 setuptools: 39.1.0-py36_0 six: 1.11.0-py36h0e22d5e_1 sqlite: 3.23.1-hf1716c9_0 tk: 8.6.7-h35a86e2_3 wheel: 0.31.0-py36_0 xz: 5.2.3-h727817e_4 zlib: 1.2.11-hf3cbc9b_2Ok, I've now stumbled across this one #407
error: [Errno 2] Could not find C/C++ file astropy/convolution/boundary_extend.(c/cpp) for Cython file astropy/convolution/boundary_extend.pyx when building extension astropy.convolution.boundary_extend. Cython must be installed to build from a git checkout.: 'astropy/convolution/boundary_extend.c'Cython is specified as a build requirement in my
recipe/meta.yml.This is actually not related to
conda build, it occures from just a./setup buildfrom the astropy src dir. Blowing away theastropy/builddir and agit cleanseem to have fixed this. I wonder if it's pulling compiler info from the built file _compiler.blahblah? If so, perhaps it shouldn't.Whoops!!! Removing the build dir from the src path seems to have fixed everything. Guess I must have done a half build at somepoint, directly using
./setup buildfrom within the src tree and just left it. Conda build obviously (now) copies the entire src tree, including this, and probbaly only carries on from where the previous build left off, or something else that mangles the build - perhaps only its presence.Anyhow, all sorted, going to slowly skulk away now without anyone noticing, bye... shhhhhhhh
#6424 doesn't seem to have been fixed. @drdavella
Building 9a5e356 using
conda-buildand, effectively, the default conda-forge/astropy-feedstock, the followinng error (and similar for other test packages) is generated.In
recipe/meta.ymlI only changed thesource(pointing to local path to astropy clone) and addedjinja2andcythonto build requirements.Furthermore, it builds! The test mechanism isn't aware that the feedstock tests failed to even run.
The package can be installed,
conda install --use-local astropy, and the same errors are given when callingastropy.test()from python.The above is true for all test packages.
The
conda-builddependencies installed are:When using the exact feedstock (nothing altered) they are the same as above.