Philipp von Weitershausen wrote:

Baiju M wrote:

Log message for revision 72757:
  Make sure including in configuration works.

I fail to see the point of this test. This is a doctest, I would at least expect one line explaining why we need to test loading

Also, why does this need to be in zope.configuration? It seems wrong introducing this kind of dependency, evne if it's just for testing.

I realise that the place of that test is wrong, I will revert it.

I do realize that zope.configuration now has some code to support backward compatibility of <include package="" />. *This* would be something worthwile testing, I suppose, though perhaps in a way that would prevent actually loading all of the configuration (perhaps toggling the execution of ZCML actions to false would be a solution).

This test loads a fair bit of stuff, pretty close to a functional test setup. This has to be torn down (which is pretty hard to do, hence the functional layers so far don't implement tearDown).

Since xmlconfig.include method only parse and not execute ZCML, toggling the execution of ZCML actions to false is not required.

+    >>> try:
+ ... xmlconfig.include(context, 'file_not_exists.zcml',
+    ... except IOError, msg:
+    ...     'OK'
+    'OK'
+    """

I think the preferred way to test for exceptions in doctests is:

>>> xmlconfig.include(context, 'file_not_exists.zcml',
  Traceback (most recent call last):
  IOError: ...

Thanks, I will change the other tests like that.

Baiju M

Zope3-dev mailing list

Reply via email to