On Monday 11 April 2016 20:13:02 Florian Apolloner wrote:
> On Monday, April 11, 2016 at 6:57:46 PM UTC+2, Tim Graham wrote:
> > I cannot think of much we could do besides monkey-patching the
> > custom-user model.
>
> I would not call checking and rewriting the class in __new__
> monkey-patching, but…
I think this is what we must do in order to keep backwards-compatibility while
resolving the issue.
If we don't resolve the issue, people will just keep using the method as if it
were a property, opening themselves up in various ways;
If we just make it a property in the builtin classes, custom user classes will
re-introduce the issue.
We can handle it in two ways, as far as I can see:
A) What Florian alluded to above -- give AbstractBaseUser a metaclass which
replaces the methods with Aymeric's "CallableBool"s
B) Implement in AbstractBaseUser a __getattribute__ which, essentially, does
the same thing when asked for the methods.
When idea which I discarded was to "force" the attributes to be used as
methods -- to use a callable which raises an exception when evaluated as
boolean. But this replacement suffers exactly the same problems as the
CallableBool replacement, unless also supported by "monkey patching"
techniques, and a CallableBool is much more user-friendly.
Have fun,
Shai.
