[Python-Dev] built-in Python test runner (was: Python Language Summit at PyCon: Agenda) (original) (raw)

Daniel Holth dholth at gmail.com
Tue Mar 5 21:20:24 CET 2013

On Tue, Mar 5, 2013 at 2:49 PM, Barry Warsaw <barry at python.org> wrote:

On Mar 05, 2013, at 02:11 AM, Donald Stufft wrote:

Doesn't setuptools/distribute already have a setup.py test command? That seems like the easiest way forward? Yes, and in theory it can make python setup.py test work well. But there are lots of little details (such as API differences for ensuring that doctests run, "additional tests" discovery, etc.) that make this often not work so well in practice. Some of that is social and some of it is technical. I still claim that including test suite information in a package's metadata would be a win, but maybe that's just too much to hope for right now.

It would be a win, but "parsing the metadata" is just not what happens right now, let alone writing anything about which and where the modules are defined in the sdist. We can barely install packages by using the dependency metadata from PKG-INFO; pip always re-generates it from "setup.py egg_info".

Your testing metadata prototype would only have to write two lines to the metadata instead of one a-la: Extension: flufl; flufl/test_suite: nose.collector; document the extension; write some tool to actually parse the metadata and invoke the tests; it may become a core feature in the next version, or having a monolithic specification may become less important.

Thanks,

Daniel Holth

More information about the Python-Dev mailing list