dulwich-users team mailing list archive

[Jelmer Vernooij <jelmer@xxxxxxxxx>]

On Sun, Dec 26, 2010 at 04:23:54PM -0500, Augie Fackler wrote:
> >> We can codify that in setup.py and make
> >> it the One True Test System (and depending on unittest2 for old
> >> Pythons means that we can continue using the normal SkipTest
> >> approach *and* dispense with the madness in test/__init__ that tries
> >> to get the "right" SkipTest. I'm happy to take on this work if
> >> nobody objects.
> > I can live with unittest2 as default fallback test runner but
> > would like to keep the support for using testtools without the need to
> > install unittest2. What do you think about the attached patch?
> Should probably also list unittest2 in the test_requires stanza of the
> setup() call in setup.py if sys.version < (2,7).
That's a good point.

> Also: testtools can't handle normal unittest.TestCase subclasses? It feels
> like testtools should be adapting to unittest2 automatically, not users of
> testtools having to patch their code to deal with test skipping.
It can, but if we're using a non-standard TestSkip class it will need to know
about that - previously we could be using nose's SkipTest for example.

I'll file a bug in testtools about its test runner being able to recognize the
SkipTest exception from unittest2, that seems like a sensible thing to do.

We don't use much of the other functionality testtools offers.

> > Upstream Python appears to have gone with SkipTest as the name of the
> > exception while we use TestSkipped. I don't have any particular
> > preference for what we should use.
> I guess it'd be nice to use the conventional name, but whatever. Not worth the churn.
Same here. I'd be happy to push patches that fix this.

Cheers,

Jelmer