Hi Maxim! Maxim Cournoyer aliandika: > Hi, > > Munyoki Kilyungi writes: > >> "jgart" aliandika: >> >>> Hi Guixers, >>> >>> I was recently reviewing a patch by a contributor and was discussing the notion of preferring upstream to PyPI when tests are missing on PyPI. >>> >> What's the policy on this. Shouldn't we prefer >> upstream over PyPI? Sometimes, PyPI versions lag >> behind upstream versions. > > There's no written policy about it at this time, but I think it's common > knowledge in Guix that we try hard to: > > 1. Unbundle dependencies (that one is documented) > 2. Run test suites (that one isn't) > > We could hint at this globally in the contributing section, and perhaps > expound with more specific details in the Python Modules contributing > section, mentioning that when the PyPI source archive lacks tests, > fetching from the git repository should be preferred. > > Another thing we could do is default to fetch from git for the Python > importer. > Thanks for this info. I'm dedicating an hour of my day from work to try to contrib to the eco-system with Jgarte's help. What's, if I may naively ask, the process for documenting this knowledge? I can take a stub at trying to document this and asking for reviews. >>> WDYT if we also signalled a special condition when there are missing tests in PyPI? >>> >> I posit that this should be the responsibility of >> the contributor. Adding extra glue can add >> unnecessary complexity. > > I think an error would help catching this situation; the > gnu-build-system would fail running 'make check' when there is no check > target for example, but in the Python world, running 'python setup.py > test' often succeeds even 0 tests were run, which is easy to overlook. > I'm not sure how easy it'd be to check for missing tests though. > TIL'ed. -- (Life is like a pencil that will surely run out, but will leave the beautiful writing of life.) (D4F09EB110177E03C28E2FE1F5BBAE1E0392253F (hkp://keys.openpgp.org))