GTK: make more use of gtk_signal convenience macros/signal registrar.

GTK: make more use of gtk_signal convenience macros/signal registrar.

BUG=none
TEST=none

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=45850

[Elliot Glaysher, 2010-04-28 16:39:12]

http://codereview.chromium.org/1708013/diff/1/2
File chrome/browser/gtk/hover_controller_gtk.cc (left):

http://codereview.chromium.org/1708013/diff/1/2#oldcode60
chrome/browser/gtk/hover_controller_gtk.cc:60: this);
I'm not so sure that these are equivalent. Previously, you were disconnecting
from button_'s OnButtonDestroyThunk before you called g_object_unref(button_).
Now that will happen afterwards. And why disconnect here in the destroy thunk at
all?

[Evan Stade, 2010-04-28 19:00:51]

http://codereview.chromium.org/1708013/diff/1/2
File chrome/browser/gtk/hover_controller_gtk.cc (left):

http://codereview.chromium.org/1708013/diff/1/2#oldcode60
chrome/browser/gtk/hover_controller_gtk.cc:60: this);
On 2010/04/28 16:39:13, Elliot Glaysher wrote:
> I'm not so sure that these are equivalent. Previously, you were disconnecting
> from button_'s OnButtonDestroyThunk before you called g_object_unref(button_).

it's not equivalent, but I believe the new code to be correct. Also note that
the other two signal handlers were not being disconnected when
HoverControllerGtk was destructed, which is a bug, and is exactly what the
signal registrar is designed to prevent.

> Now that will happen afterwards. And why disconnect here in the destroy thunk
at
> all?

yea, not much point since the signal handler will disconnect itself on destroy.

[Elliot Glaysher, 2010-04-28 19:09:56]

OK, LGTM then.

On Wed, Apr 28, 2010 at 12:00 PM,  <estade@chromium.org> wrote:
>
> http://codereview.chromium.org/1708013/diff/1/2
> File chrome/browser/gtk/hover_controller_gtk.cc (left):
>
> http://codereview.chromium.org/1708013/diff/1/2#oldcode60
> chrome/browser/gtk/hover_controller_gtk.cc:60: this);
> On 2010/04/28 16:39:13, Elliot Glaysher wrote:
>>
>> I'm not so sure that these are equivalent. Previously, you were
>
> disconnecting
>>
>> from button_'s OnButtonDestroyThunk before you called
>
> g_object_unref(button_).
>
> it's not equivalent, but I believe the new code to be correct. Also note
> that the other two signal handlers were not being disconnected when
> HoverControllerGtk was destructed, which is a bug, and is exactly what
> the signal registrar is designed to prevent.
>
>> Now that will happen afterwards. And why disconnect here in the
>
> destroy thunk at
>>
>> all?
>
> yea, not much point since the signal handler will disconnect itself on
> destroy.
>
> http://codereview.chromium.org/1708013/show
>