launchpad-dev team mailing list archive

[Abel Deuring <abel.deuring@xxxxxxxxxxxxx>]

On 26.07.2010 11:29, Robert Collins wrote:
> On Mon, Jul 26, 2010 at 10:10 AM, Abel Deuring
> <abel.deuring@xxxxxxxxxxxxx> wrote:
> 
>> 1. I don't know why we raised errors for non-silenced warnings. In
>> general, we should aim to get rid of all warnings in our code/tests. I
>> would even consider the warnings introduced by my branch to be errors --
>> the point is that they occur far too often to be fixed in a single
>> branch, so I opted instead to annoy everbody who looks at the output of
>> affected tests ;)
> 
> For future reference, see the patch James W put together to attach
> oopses to tests; a similar approach would have let you gather all the
> items without them showing up : we could then ratchet it down as a
> form of lint without having the UI exploding in our faces.
> 
> Separately, I'm a little concerned we may spend a lot of time churning
> between users to setup things now - and that that will have
> performance implications for the test suite.

well, if this is a concern, simply calling removeSecurityProxy() should
be fine in a setUp() method -- provided that the need to do that does
not indicate a permission problem in the "real world". Actually, I used
removeSecurityProxy(obj) in LPObjectFactory.makeSomething(), if this
method failed because obj is returned by LPObjectFactory.makeOther() and
makeOther() now returns a proxied object.

> 
> Lastly, and here I expose my ignorance of some subtleties in zope - I
> thought security proxies only lived between view and model objects,
> not between model objects?

right. But many tests work like views (in the sense that login() calls
are necessary), so LPObjectFactory should provide them with proxied
objects. (Or did I miss your point?)

Abel