>  Looks good! I wonder if Peter could look at/review the patch? I'll try to
> do it myself too but I don't know when can I manage it and, anyhow, 2 heads
> (4 eyes?) are better than one (2).
>
Code review for newevent15.patch

src/gtk/spinctrl.cpp(32): Will the TODO be done?

src/gtk/spinbutt.cpp(30): Will the TODO be done?

Did I overlook them or why haven't these events been changed as well?
src/common/event.cpp(212-223,282,289,297-299): 

src/common/event.cpp(926): Why was the check 'm_fn != 0' changed? Even though 
m_fn is now a wxEventFunctor pointer, it should still be valid, but I'm not 
sure whether this is also true for the new check.

src/common/event.cpp(1271): The comment: "no need for an extra virtual 
function call" should be removed, since it is not true anymore.

src/common/event.cpp(1460): Wouldn't a parameter by reference be better here?

include/wx/event.h(97): I wouldn't use the term 'modern events' here. I used 
it in the prototype only to make it more clear, what events are legacy and 
what are the improved (modern) events. I would write:
// macros to create an event type depending on whether type safe events are 
enabled. 

include/wx/event.h(165): I would move the destructor to event.cpp. I think  
some compilers have difficulties creating the vtbl if the destructor is 
defined inline, but I'm not sure.

include/wx/event.h(380): Even though I understand the reasoning from Brian, 
I'm still not convinced we need the wxEventConnection. I have the feeling it 
makes in unnecessary complex to introduce a proxy object.

include/wx/event.h(387,399): Wouldn't the compiler generated copy constructor 
and assignment operator be sufficient?

include/wx/event.h(2518): Shouldn't that be wx_const_cast() and not the plain 
c++ const_cast<>() ?

include/wx/event.h(2534): Typo: Should read "associated" and not "asosciated". 

include/wx/event.h(2816): Since the 'Connect' method is private this comment 
isn't correct.