[Mac] Line up omnibox popup under field.

[Mac] Line up omnibox popup under field.

Also line up the icons.  Spacing can be adjusted later.

BUG=37865
TEST=Popup edges line up under field edges.
TEST=Popup icons and text should line up under field icon and text.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=43357

[Scott Hess - ex-Googler, 2010-04-01 00:22:19]

Split this out of a CL which moves the star button into the field to make that
CL less obscure.

[rohitrao (ping after 24h), 2010-04-01 01:15:19]

Looks like we have a unittest to make sure the popup is positioned correctly. 
Which it's not anymore.  Or maybe the unittest is wrong now :)

Looking at code now.

[Scott Hess - ex-Googler, 2010-04-01 01:15:57]

Crap.

On Wed, Mar 31, 2010 at 6:15 PM,  <rohitrao@chromium.org> wrote:
> Looks like we have a unittest to make sure the popup is positioned
> correctly.
> Which it's not anymore.  Or maybe the unittest is wrong now :)
>
> Looking at code now.
>
> http://codereview.chromium.org/1608001
>

[Scott Hess - ex-Googler, 2010-04-01 01:17:45]

Yeah, unit test is wrong, now.  I'll have to get it in the morning.

On 2010/04/01 01:15:57, shess wrote:
> Crap.
> 
> On Wed, Mar 31, 2010 at 6:15 PM,  <mailto:rohitrao@chromium.org> wrote:
> > Looks like we have a unittest to make sure the popup is positioned
> > correctly.
> > Which it's not anymore. &nbsp;Or maybe the unittest is wrong now :)
> >
> > Looking at code now.
> >
> > http://codereview.chromium.org/1608001
> >
>

[rohitrao (ping after 24h), 2010-04-01 01:44:07]

Looks fine other than the broken test, although I'm a tiny bit worried that we
don't understand the top/bottom confusion with the rounded rect.

http://codereview.chromium.org/1608001/diff/1/4
File chrome/browser/cocoa/toolbar_controller.mm (right):

http://codereview.chromium.org/1608001/diff/1/4#newcode855
chrome/browser/cocoa/toolbar_controller.mm:855: const int
kLocationStackEdgeWidth = 1;
Why did this change?

[Scott Hess - ex-Googler, 2010-04-01 03:01:06]

On 2010/04/01 01:44:07, rohitrao wrote:
> Looks fine other than the broken test, although I'm a tiny bit worried that we
> don't understand the top/bottom confusion with the rounded rect.

This is an NSMatrix subclass, I'm pretty sure -isFlipped returns YES, so it
makes sense to be upside-down.

Changes uploaded, hopefully the tests will pass this time...

http://codereview.chromium.org/1608001/diff/1/4
File chrome/browser/cocoa/toolbar_controller.mm (right):

http://codereview.chromium.org/1608001/diff/1/4#newcode855
chrome/browser/cocoa/toolbar_controller.mm:855: const int
kLocationStackEdgeWidth = 1;
On 2010/04/01 01:44:07, rohitrao wrote:
> Why did this change?

Previously, I believed that it was |2| because of something to do with the
rounded endcaps - but I'm pretty sure that the value was just chosen by
experiment.

This time, I believe it's more sensible.  The outline of the window, which is
outside the frame, needs to match the outline of the field, which is included in
locationFrame below.  I've modified the comment to that end.

Though I will be honest - it looked wrong, I used Pixie.app to verify that it
was one pixel wrong, and subtracted that pixel.

[rohitrao (ping after 24h), 2010-04-01 03:19:18]

LGTM