Merge lp://staging/~mbp/launchpad/feature-modulus into lp://staging/launchpad

This doesn't do anything much towards getting closed loop metrics from it. We could do something either in js or the webapp or both. Probably the webapp should just always log all the active features.

https://bugs.launchpad.net/launchpad/+bug/666544 (which I didn't have access to when i wrote this offline) suggests that I need to test it for anonymous requests, and secondly that we can do a follow-on thing, or a separate scope, that activates things for a percentage of sessions, which will help for anon users.

Very nice :)

[1]

+ def __init__(self, get_person):
+ self._get_person = get_person
+ self._person = None

I don't think the _person attribute is used anywhere.

[2]

+ def test_user_modulus(self):
+ person = self.factory.makePerson()

This is somewhere a test double would work well. Something like:

    class person:
        id = 123

would mean that the test could run without a layer specified (and the
base class would have to be TestCase instead of -WithFactory).

I guess it might be better to leave it as it is though, if new tests
will be added in subsequent branches.

> Very nice :)
>
>
> [1]
>
> + def __init__(self, get_person):
> + self._get_person = get_person
> + self._person = None
>
> I don't think the _person attribute is used anywhere.

thanks.

>
>
> [2]
>
> + def test_user_modulus(self):
> + person = self.factory.makePerson()
>
> This is somewhere a test double would work well. Something like:
>
> class person:
> id = 123
>
> would mean that the test could run without a layer specified (and the
> base class would have to be TestCase instead of -WithFactory).
>
> I guess it might be better to leave it as it is though, if new tests
> will be added in subsequent branches.

I see what you mean. Probably I should also have a larger test that the scope is correctly detected from the request, by doing a request with a person logged in.