[Python-projects] pylint-0.13.1 (minor) test issues

Sylvain Thénault sylvain.thenault at logilab.fr
Mon Mar 5 09:52:44 CET 2007

On Friday 02 March à 23:33, Marien Zwart wrote:
> Hi,

Hi Marien,
> Thanks for the 0.13.1 tarball. Running the tests before bumping
> gentoo's ebuild I ran into a few minor problems, most of them in the
> tests, not pylint itself.
> Everything is with pylint 0.13.1, astng 0.17.0, common 0.21.2.
> setup.py does not install the test/rpythonmessages dir. Adding it to
> include_dirs in __pkginfo__.py fixes this. Obviously not a big deal
> since I do not think many users will want to install and then run the
> tests on the installed version, but since test/messages *is* installed
> it would be nice to be consistent here.

> The rpythoninput/func_unsupported_protocol.py test fails: the line
> "I0011: 1: Locally disabling E1202" is not printed. Have not figured
> out why.

this message shouldn't be in the messages file, fixed
> The test_checker_dep_graphs in test_import_graph fails. I think this
> is because the environment I run the tests in differs from yours: in
> checkers.imports.ImportsChecker.report_dependencies_graph
> self.stats['dependencies'] is an empty dict, and if I change the test
> to call l.check('pylint') instead of l.check('input') the test passes.
> So it looks like the "input" it is running pylint on has no external
> deps, causing it to skip the dotfile generation.

hum, weird anyway. I'll investigate this later.
> Finally a problem in pylint itself instead of the tests: "pylint
> --rpython-mode test/rpythoninput/func_unsupported_protocol.py" prints
> the usage info, as if I had typed pylint with no arguments.
> preprocess_options removes the argument after rpython-mode from argv
> and passes it to cb_rpython_mode as "value", which ignores it. "pylint
> --rpython-mode please test/rpythoninput/func_unsupported_protocol.py"
> works :)

> (before I forget: there is no need to keep runtests.py or something
> like it around just for gentoo's ebuild. I will just adapt the ebuild
> if the preferred test runner changes. Would like to keep running the
> tests from the ebuild to catch any obvious mistakes though.)

fine. I love to get testing reports like yours when I publish a new
release, so please keep them running :)

Sylvain Thénault                               LOGILAB, Paris (France)
Formations Python, Zope, Plone, Debian:  http://www.logilab.fr/formations
Développement logiciel sur mesure:       http://www.logilab.fr/services
Python et calcul scientifique:           http://www.logilab.fr/science

More information about the Python-Projects mailing list