Fix memory leaks in the balloon controller unit test by loading the NIB in th...

Fix memory leaks in the balloon controller unit test by loading the NIB in the correct test-friendly way, which allows the windows to be closed correctly and the cleanup code to run as expected.

BUG=49590
TEST=unittests


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=54018

[Nico, 2010-07-28 18:58:40]

LG

thanks for taking the time to track this down

http://codereview.chromium.org/2825073/diff/1/3
File chrome/browser/cocoa/notifications/balloon_controller_unittest.mm (right):

http://codereview.chromium.org/2825073/diff/1/3#newcode73
chrome/browser/cocoa/notifications/balloon_controller_unittest.mm:73:
BalloonController* controller = [BalloonController alloc];
move the init call to here

http://codereview.chromium.org/2825073/diff/1/3#newcode94
chrome/browser/cocoa/notifications/balloon_controller_unittest.mm:94:
BalloonController* controller = [BalloonController alloc];
same

[John Gregg, 2010-07-28 19:02:11]

On 2010/07/28 18:58:40, Nico wrote:
> LG
> 
> thanks for taking the time to track this down
> 
> http://codereview.chromium.org/2825073/diff/1/3
> File chrome/browser/cocoa/notifications/balloon_controller_unittest.mm
(right):
> 
> http://codereview.chromium.org/2825073/diff/1/3#newcode73
> chrome/browser/cocoa/notifications/balloon_controller_unittest.mm:73:
> BalloonController* controller = [BalloonController alloc];
> move the init call to here
> 
> http://codereview.chromium.org/2825073/diff/1/3#newcode94
> chrome/browser/cocoa/notifications/balloon_controller_unittest.mm:94:
> BalloonController* controller = [BalloonController alloc];
> same

So now that I remember this code, the reason the alloc and init are split like
that is that I need to mock out a method that init is going to use.  Is there
any other way to do that?  I can also add a comment to that effect.

[Nico, 2010-07-28 19:04:47]

On Wed, Jul 28, 2010 at 12:02 PM,  <johnnyg@chromium.org> wrote:
> On 2010/07/28 18:58:40, Nico wrote:
>>
>> LG
>
>> thanks for taking the time to track this down
>
>> http://codereview.chromium.org/2825073/diff/1/3
>> File chrome/browser/cocoa/notifications/balloon_controller_unittest.mm
>
> (right):
>
>> http://codereview.chromium.org/2825073/diff/1/3#newcode73
>> chrome/browser/cocoa/notifications/balloon_controller_unittest.mm:73:
>> BalloonController* controller = [BalloonController alloc];
>> move the init call to here
>
>> http://codereview.chromium.org/2825073/diff/1/3#newcode94
>> chrome/browser/cocoa/notifications/balloon_controller_unittest.mm:94:
>> BalloonController* controller = [BalloonController alloc];
>> same
>
> So now that I remember this code, the reason the alloc and init are split
> like
> that is that I need to mock out a method that init is going to use.  Is
> there
> any other way to do that?  I can also add a comment to that effect.

manual subclass + override might be cleaner?

>
> http://codereview.chromium.org/2825073/show
>

[John Gregg, 2010-07-28 20:24:18]

On 2010/07/28 19:04:47, Nico wrote:
> On Wed, Jul 28, 2010 at 12:02 PM,  <mailto:johnnyg@chromium.org> wrote:
> > On 2010/07/28 18:58:40, Nico wrote:
> >>
> >> LG
> >
> >> thanks for taking the time to track this down
> >
> >> http://codereview.chromium.org/2825073/diff/1/3
> >> File chrome/browser/cocoa/notifications/balloon_controller_unittest.mm
> >
> > (right):
> >
> >> http://codereview.chromium.org/2825073/diff/1/3#newcode73
> >> chrome/browser/cocoa/notifications/balloon_controller_unittest.mm:73:
> >> BalloonController* controller = [BalloonController alloc];
> >> move the init call to here
> >
> >> http://codereview.chromium.org/2825073/diff/1/3#newcode94
> >> chrome/browser/cocoa/notifications/balloon_controller_unittest.mm:94:
> >> BalloonController* controller = [BalloonController alloc];
> >> same
> >
> > So now that I remember this code, the reason the alloc and init are split
> > like
> > that is that I need to mock out a method that init is going to use. &nbsp;Is
> > there
> > any other way to do that? &nbsp;I can also add a comment to that effect.
> 
> manual subclass + override might be cleaner?
> 
> >
> > http://codereview.chromium.org/2825073/show
> >
> 

Okay please take another look.

[Nico, 2010-07-28 20:26:34]

LG++