Scrap WNDCLASSEX.hCursor, update GetCursorForPoint, etc.

Scrap WNDCLASSEX.hCursor, update GetCursorForPoint.
Default to Win arrow in View::GetCursorForPoint & RootView::UpdateCursor.
Simplify WidgetWin::SetCursor, avoid sending NULL.
Only SetCuror on client events (DWM handles non-client).

RIP TextButtonWithHandCursorOver (r57652 through r64531).
RIP WindowWin::InitClass & |resize_cursors_|.
Add OVERRIDE specifiers liberally.

BUG=35356
TEST=Cursor styles shown in Win & Linux toolkit_views.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=83123

[msw, 2011-04-26 22:08:39]

Hey Peter, can you review this? Should I include anyone else?

It might be beneficial to create a cross-platform views::Cursor in subsequent
CL; thoughts?

[Peter Kasting, 2011-04-27 00:22:12]

LGTM

http://codereview.chromium.org/6880201/diff/2001/ui/base/win/window_impl.cc
File ui/base/win/window_impl.cc (right):

http://codereview.chromium.org/6880201/diff/2001/ui/base/win/window_impl.cc#n...
ui/base/win/window_impl.cc:201: WNDCLASSEX class_ex;
Nit: Does it make sense to use array initializer style (= {...}) here?

http://codereview.chromium.org/6880201/diff/2001/views/controls/link.h
File views/controls/link.h (right):

http://codereview.chromium.org/6880201/diff/2001/views/controls/link.h#newcode47
views/controls/link.h:47: virtual void SetEnabled(bool flag) OVERRIDE;
Nit: |flag | is better than |f|.  |enabled| might be even better.

http://codereview.chromium.org/6880201/diff/2001/views/view.h
File views/view.h (left):

http://codereview.chromium.org/6880201/diff/2001/views/view.h#oldcode586
views/view.h:586: // Return the cursor that should be used for this view or NULL
if
Nit: This still can return NULL in some cases.  Should this sentence have been
left as-is?

http://codereview.chromium.org/6880201/diff/2001/views/view.h
File views/view.h (right):

http://codereview.chromium.org/6880201/diff/2001/views/view.h#newcode590
views/view.h:590: //shared resource but Gtk destroys the returned cursor after
setting it.
Nit: Initial space; comma after "resource"

[msw, 2011-04-27 01:58:03]

Thanks, comments addressed; landing after clean try.

http://codereview.chromium.org/6880201/diff/2001/ui/base/win/window_impl.cc
File ui/base/win/window_impl.cc (right):

http://codereview.chromium.org/6880201/diff/2001/ui/base/win/window_impl.cc#n...
ui/base/win/window_impl.cc:201: WNDCLASSEX class_ex;
On 2011/04/27 00:22:12, Peter Kasting wrote:
> Nit: Does it make sense to use array initializer style (= {...}) here?

Done. Seems like no comments is the most comment pattern.

http://codereview.chromium.org/6880201/diff/2001/views/controls/link.h
File views/controls/link.h (right):

http://codereview.chromium.org/6880201/diff/2001/views/controls/link.h#newcode47
views/controls/link.h:47: virtual void SetEnabled(bool flag) OVERRIDE;
On 2011/04/27 00:22:12, Peter Kasting wrote:
> Nit: |flag | is better than |f|.  |enabled| might be even better.

I'm following the base definition; I think that's optimal, short of changing all
occurrences to |enabled|.

http://codereview.chromium.org/6880201/diff/2001/views/view.h
File views/view.h (left):

http://codereview.chromium.org/6880201/diff/2001/views/view.h#oldcode586
views/view.h:586: // Return the cursor that should be used for this view or NULL
if
On 2011/04/27 00:22:12, Peter Kasting wrote:
> Nit: This still can return NULL in some cases.  Should this sentence have been
> left as-is?

With native cursor types, NULL should be used explicitly to mean:
WIN: "the cursor is removed from the screen"
http://msdn.microsoft.com/en-us/library/ms648393(v=vs.85).aspx
GDK: "use the cursor of its parent window"
http://developer.gnome.org/gdk/stable/gdk-Windows.html#gdk-window-set-cursor

I nixed the behavior that replaced NULL with default/|previous_cursor_| from
WidgetWin, and WidgetGtk seemingly already uses the native behavior.

We ought to eventually create a cross platform cursor type/enum.

http://codereview.chromium.org/6880201/diff/2001/views/view.h
File views/view.h (right):

http://codereview.chromium.org/6880201/diff/2001/views/view.h#newcode590
views/view.h:590: //shared resource but Gtk destroys the returned cursor after
setting it.
On 2011/04/27 00:22:12, Peter Kasting wrote:
> Nit: Initial space; comma after "resource"

Done.