[Zope-CMF] CMF 2 test errors against zope 2.9 branch and trunk

[Paul Winkler]

On Sat, Apr 08, 2006 at 02:37:32AM +0200, Stefan H. Holek wrote:
> FWIW I always get this one with 2.9. Didn't investigate further.
> 
> Error in test test_reindexObjectSecurity_missing_noraise  
> (CMFCore.tests.test_CMFCatalogAware.CMFCatalogAwareTests)
> Traceback (most recent call last):
>   File "/usr/local/python2.4/lib/python2.4/unittest.py", line 260,  
> in run
>     testMethod()
>   File "/home/stefan/autotest/temp/python24-zope29-cmf20/Products/ 
> CMFCore/tests/test_CMFCatalogAware.py", line 178, in  
> test_reindexObjectSecurity_missing_noraise
>     self.assertEqual( len(self.logged), 1 ) # logging because no raise
> TypeError: len() of unsized object

I've seen that too, I thought there were some issues with LogInterceptor
which I mailed Tres about privately - although Tres, now that I've
looked at the code excercised by the test above, it uses
the root logger so it *can't* be the same issue I mentioned to you, so
I'm back to being baffled. Maybe anotehr "installation brokenness"
as Jens suggests?
 
> On 8. Apr 2006, at 01:10, Jens Vagelpohl wrote:
> >On 7 Apr 2006, at 22:09, Paul Winkler wrote:
> >
> >>FYI, presumably some folks are already aware of this, but there
> >>are currently some errors when running CMF 2.0 tests against  
> >>either the
> >>zope 2.9 branch or zope 2 trunk, as follows (I have not tried all
> >>combinations):
> >
> >The tests run fine for me (well, they did a few days ago). It was  
> >mentioned on this list that Zope 2.9 installation brokenness can  
> >cause these symptoms. If it is not installed correctly you will see  
> >errors. Unfortunately, installing Zope 2.9 from anything other than  
> >a tarball is basically a crapshoot at this point.

Hmm. So from a checkkout, "make install" doesn't work at all, and "make
instance" is apparently not reliable for testing products?  It *mostly*
works, with consistent but confusing errors :-p

Testing sucks. 

-- 

Paul Winkler
http://www.slinkp.com