[Product-Developers] Thinking about CMF site fixture/layer for plone.testing

David Glick davidglick at groundwire.org
Sat May 12 17:45:27 UTC 2012


On 5/10/12 9:24 AM, Sean Upton wrote:
> I'm using a homegrown CMF site fixture/layer with plone.testing for
> testing add-ons that require CMF but not Plone (and possibly also are
> non-GPL e.g. MIT/BSD/ZPL type licenses).  What I have done thus far is
> minimal -- just the tiny bits I need to test: create a site class
> subclassing Products.CMFCore.PortalObject.PoralObjectBase, and add a
> portal_catalog to that site, and this runs inside a layer based on
> plone.testing.z2.STARTUP.
>
> This seems useful to me because it gives me a way of resolving items
> in a site via a catalog, and it gives me a persistent component
> registry to test against.  It also is much faster to set up and tear
> down than a fully-blown Plone site, which is advantageous for packages
> that have minimal coupling with Plone.
>
> Is this something that might (eventually) be useful to others as an
> added layer in plone.testing itself (assuming another optional extra
> called 'cmf', and a layer resource called 'site')?
Why not 'portal', since that's what's used in the Plone fixture, 
CMFTestCase, and PortalTestCase in Zope2's Testing package?
> If so, what OOTB CMF tools and fixtures would you want in a CMF site
> fixture for testing?
>
> Also, is there any point in using / requiring CMFDefault if a fixture
> based on CMFCore will suffice?  My goal would be only adding extra
> dependency on CMFCore only.
>
+1 for not requiring CMFDefault, and for the proposal in general.

David


----------		
David Glick
 Web Developer
 davidglick at groundwire.org
 206.286.1235x32

Are you engaging? Find out! Use our free engagement benchmarking tool.

http://groundwire.org/labs/engagement-strategy/diy-benchmarking-survey




More information about the Product-Developers mailing list