I am a !#4@head ;-) Forgot the else: clause in the code I posted. Corrected code is below. T On Thu, Aug 13, 2009 at 11:30 AM, Tim Hoffman <timh@zute.net> wrote:

Hi Folks This is a quick and dirty change and to be honest the main reason I am posting it here is so there is a searchable record for anyone else who wants to use zope.deferredimport under google app engine. (With some content models zope.deferredimport is a great way of dealing with cyclic dependancies and possibly improving startup times - see bottom of this email.)

Below is a minimal set of changes to zope.proxy.__init__ that is required so that zope.deferredimport can be used. It does mean you need peak.utils to implement a basic proxy.

incorrect code shown, Should have read try: from zope.proxy._zope_proxy_proxy import * from zope.proxy._zope_proxy_proxy import _CAPI except NotImplementedError: pass else: from peak.util.proxies import ObjectProxy class ProxyBase(ObjectProxy): pass def getProxiedObject(obj): if hasattr(obj,'__subject__'): return obj.__subject__ else: return obj def removeAllProxies(obj): return getProxiedObject(obj) def sameProxiedObjects(obj1,obj2): if getProxiedObject(obj1) == getProxiedObject(obj2): return True else: return False

There are a couple of other zope.proxy functions from the 'c' api that I haven't looked at yet as they aren't used by zope.deferredimport. Suggestions on improving it and/or followups on the other functions would also be welcome. I have also found that deferred import can help with app engine startup times as it means you can import a module you depend on but not really import the code until it gets touched which can allow you to stagger you startup costs. (For instance edit functionality of a site could be deferred until someone actually edits something).

Hope this helps someone

Regards

Tim Hoffman