[Mac] Rewrite Omnibox field-editor frame-changing tests.

[Mac] Rewrite Omnibox field-editor frame-changing tests.

The previous tests mostly replicated
autocomplete_text_field_cell_unittest.mm cases.  Reworked them to test
that the field editor placement is the same whether
-resetFieldEditorFrameIfNeeded is doing the placement, or the regular
Cocoa focusing machinery.  That's more relevant to what that method is
doing.

[Scott Hess - ex-Googler, 2009-09-23 01:23:36]

This is preliminary to landing a change to do the editor reset by other means
(which should be more compatible with things like undo management).

[Scott Hess - ex-Googler, 2009-09-23 21:44:35]

On 2009/09/23 01:23:36, shess wrote:
> This is preliminary to landing a change to do the editor reset by other means
> (which should be more compatible with things like undo management).

Ping - this is a complete CL, changing the unit test to better test something
I'm going to change with a different CL.  Putting the other change and the unit
test change together felt like cheating.

[rohitrao (ping after 24h), 2009-09-23 21:46:52]

Yup, I started looking, but then I got distracted by perf.  I'll take
another look after I finish one more :)

On Wed, Sep 23, 2009 at 5:44 PM,  <shess@chromium.org> wrote:
> On 2009/09/23 01:23:36, shess wrote:
>>
>> This is preliminary to landing a change to do the editor reset by other
>> means
>> (which should be more compatible with things like undo management).
>
> Ping - this is a complete CL, changing the unit test to better test
> something
> I'm going to change with a different CL. =A0Putting the other change and =
the
> unit
> test change together felt like cheating.
>
> http://codereview.chromium.org/219003
>

[rohitrao (ping after 24h), 2009-09-23 22:05:35]

LGTM

http://codereview.chromium.org/219003/diff/1/2
File chrome/browser/cocoa/autocomplete_text_field_unittest.mm (right):

http://codereview.chromium.org/219003/diff/1/2#newcode70
Line 70: EXPECT_TRUE([field_.get() currentEditor]);
I want this to be an ASSERT_TRUE, but I have a feeling that won't do what I
want, since this is all inside a function.

[Scott Hess - ex-Googler, 2009-09-23 23:06:25]

thanks!

http://codereview.chromium.org/219003/diff/1/2
File chrome/browser/cocoa/autocomplete_text_field_unittest.mm (right):

http://codereview.chromium.org/219003/diff/1/2#newcode70
Line 70: EXPECT_TRUE([field_.get() currentEditor]);
On 2009/09/23 22:05:35, rohitrao wrote:
> I want this to be an ASSERT_TRUE, but I have a feeling that won't do what I
> want, since this is all inside a function.

Yes, ASSERT_TRUE won't work (it's more relevant for the -count check, because
accessing the first object throws an exception).  I tried working around that,
but it wasn't reasonable.  Making the function return void allows ASSERT_TRUE(),
but makes the helper function useless.  Putting the ASSERT higher up likewise
reduces the utility.

Now that I go look, I find that -lastObject will return the last object in the
array, or nil if empty.  That might have had potential, but I'm not sure what
[nil frame] would return ...