(Taking the liberty do so, feel free to ignore this e-mail if you don't find these good)
As is, the build time test suite was choking on py.test, so I
begun working on a small patch against the Makefile to rename
this py.test-3 and build depend on "python3-pytest <!nocheck>".
Rather than patching the Makefile, you could simply --with python3 --buildsystem=pybuild
This saves us from un-necessary delta with upstream.
However when adding a few more test dependencies[1], I hit that
wall, where registration of the q2 modules seems to be expected:
dh_auto_test
make -j6 test
make[1]: Entering directory '/<<PKGBUILDDIR>>'
py.test-3
============================= test session starts ==============================
platform linux -- Python 3.9.1rc1, pytest-4.6.11, py-1.9.0, pluggy-0.13.0
rootdir: /<<PKGBUILDDIR>>
collected 0 items / 6 errors
==================================== ERRORS ====================================
_________ ERROR collecting q2_sample_classifier/tests/test_actions.py __________
ImportError while importing test module '/<<PKGBUILDDIR>>/q2_sample_classifier/tests/test_actions.py'.
Hint: make sure your test modules/packages have valid Python names.
Traceback:
q2_sample_classifier/tests/test_actions.py:13: in <module>
from qiime2.plugins import sample_classifier
E ImportError: cannot import name 'sample_classifier' from 'qiime2.plugins' (/usr/lib/python3/dist-packages/qiime2/plugins.py)
[1] namely build time test dependencies are : python3-distutils,
q2-types and qiime.
Maybe the debian/rules will need an:
override_dh_auto_test:
# FIXME: test delayed to autopkgtest
and the package will need an autopkgtest which executes
py.test-3 on the installed module distribution, a bit similar to
what qiime and q2-types are doing at the moment?
Yes, final agenda is to execute pytest and see if it works well or not.
Hence disabling build-time test and executing pytest is good.
I've seen other packages under med-team as well which do the same thing - because build-time tests are not run-able w/o package installation.
I didn't have anything to push, I'm not sure how you would wish
to move forward on this one.
> Many thanks,
>
> Steffen (hoping ITPs also count for the calendar)
>
> We should actually think of some Xmas / New Year Promises packaging
> story, like comparing the microbiome of the present, the past and the
> future. The age-effect comparing with the past is obvious (drinking
> mostly milk, parental priming in the very early days). That "future"
> thingy is more real than what the wider public is yet aware of, for one
> there is our immune system that is changing, but there is also
> nutritional genomics so we are not completely out of control. Anybody
> feeling creative?
Sounds interesting, but I'm not sure I understood your idea? :)
+1 :-P