Implemented LazyObject class, which provides a 'lazy' class method to create instances with lazy initialization. Th… #665
Conversation
crunk1
changed the title
| self._request() | ||
|
|
||
| def _request(self): | ||
| # TODO(crunkleton@google.com): create new connection? or make |
There was a problem hiding this comment.
@crunk1 so what is the intention with the todo? to raise a separate PR later
There was a problem hiding this comment.
Yes, I need to look into this later.
|
LGTM +1 |
| pass | ||
|
|
||
| def __getattribute__(self, attr): | ||
| lazy_obj = object.__getattribute__(self, 'lazy_obj') |
There was a problem hiding this comment.
Should you prefix the lazy_obj attr with an underscore (_lazy_obj) to indicate it's intended to be a private attr?
There was a problem hiding this comment.
Also, I believe if you override __getattr__ instead of __getattribute__ then you can just check the attribute value without having to invoke __getattribute__ directly like this:
if self.lazy_obj is None:
But maybe there's something subtle here I'm missing?
There was a problem hiding this comment.
Changed LazyObject.Proxy.lazy_obj to LazyObject.Proxy._lazy_obj.
Regarding the __getattr__ vs __getattribute__ discussion, see http://stackoverflow.com/a/3278104/1436495.
My considerations:
__getattr__is only called if attribute resolution fails by normal means__getattribute__is ALWAYS called (even on bound methods); easier to handle given my next point- the
Proxymetaclass inherits fromlazy_cls(I wantProxyto be as transparent as possible as a wrapper; I want the returnedProxyinstance to have the class info forlazy_cls), meaning thatProxyinherits all methods and class attributes oflazy_cls. Therefore, weird things happen when you call methods or use class-level attributes intended for the lazy object instead of theProxy.
…e lazy class method returns a Proxy object that subclasses the target object class. Upon accessing the proxy object in any way, the object is initialized. Modified Google Compute Engine License objects, GCELicense, to be such a lazy object. This addresses https://issues.apache.org/jira/browse/LIBCLOUD-786. Tests/Verification: tox -e lint python setup.py test Added test/common/test_base.py which has LazyObjectTest
| pass | ||
|
|
||
| def __getattribute__(self, attr): | ||
| lazy_obj = object.__getattribute__(self, '_get_lazy_obj')() |
There was a problem hiding this comment.
Wrapping it in a method was nice, but I don't think you need to use __getattribute__ to call that method since it's defined in the same class here, so any invocation of this method is guaranteed to have self._get_lazy_obj() available to call.
There was a problem hiding this comment.
Since the Proxy class has __getattribute__ defined, self._get_lazy_obj() will first call Proxy.__getattribute__(self, '_get_lazy_obj') to get the bound method. That leads to an infinite recursion loop. See an example here: https://repl.it/Bavg/1.
There was a problem hiding this comment.
Oh, right, derp. I should stop doing code reviews before I get any caffeine in me.
There was a problem hiding this comment.
All good, thanks for all your help!
|
One last nitpick, but otherwise lgtm. |
|
Ok, this lgtm too and I've done some additional testing. The Merging now. |
|
This has now been merged: 81a8dbc |
…e lazy class method returns a Proxy object that subclasses the target object class. Upon accessing the proxy object in any way, the object is initialized.
Modified Google Compute Engine License objects, GCELicense, to be such a lazy object. This addresses https://issues.apache.org/jira/browse/LIBCLOUD-786.
Tests/Verification:
tox -e lint
python setup.py test
Added test/common/test_base.py which has LazyObjectTest