I meant that somehow the Windows events were connected to user defined functions using boost::bind or something like that.

Are you sure you're talking about Windows events? Events are just synchronization objects, they are not connected to any code. http://msdn.microsoft.com/en-us/library/ms682396(v=vs.85).aspx

Hi John, Yes, exactly. Windows messages (I want to avoid having to write long case statements in message loops) - that's what I mean. I'd be thankful if anyone could point me to the right direction. Thank you very much, Asif On Thu, Mar 24, 2011 at 3:26 PM, John Drescher wrote:

On Thu, Mar 24, 2011 at 6:13 AM, Igor R wrote:

...

...

I meant that somehow the Windows events were connected to user defined functions using boost::bind or something like that.

Are you sure you're talking about Windows events? Events are just synchronization objects, they are not connected to any code. http://msdn.microsoft.com/en-us/library/ms682396(v=vs.85).aspx

I think he is talking about the windows message loop.

http://en.wikipedia.org/wiki/Message_loop_in_Microsoft_Windows

John _______________________________________________ Boost-users mailing list Boost-users@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

Hi Igor, Thank you very much. The thing that I don't understand is that how folks such as those from SmartWin++ are able to seemlessly connect functions with Windows messages without using a case statement in a message loop. An example is at the following page: http://smartwin.sourceforge.net/getting_started.php Using your approach, I need to use a case statement inside a message loop. That's what I don't want to do. I am familiar with CRTP that SmartWin++ folks seem to have used as deriving new classes is very easy and the resulting code is very clean. So, I can at least imagine the design behind this library. But I wonder if they have written such case statements in message loops under the hood for every single GUI element. Someone on Boost Users mailing list did provide a solution to this a few months back (I do recall but cannot seem to find that post - I starred it in my thunderbird inbox but I have lost my inbox backup - I can scan the whole boost users list but I have limited time available to accomplish what I want). Best regards, Asif On Thu, Mar 24, 2011 at 4:02 PM, Igor R wrote:

...

Yes, exactly. Windows messages (I want to avoid having to write long case statements in message loops) - that's what I mean. I'd be thankful if anyone could point me to the right direction.

// pseudo-code!

// Lets say it's a botton; switch(message) { case WM_LBUTTONDBLCLK: onDoubleClick(message); break; //... };

// member function void onDoubleClick(Message message) { doSomething(message); onDoubleClick_(arg1, arg2); }

// member boost::signals2::signal onDoubleClick_;

Now multiple listeners of your button can connect the signal:

struct Handler { void onDoubleClick(Arg1 a1, Arg2 a2); }; button->onDoubleClick.connect(boost::bind(&Handler::onDoubleClick, handler, _1, _2));

This's only a sketch, in real code hou have to care about proper encapsulation, appropriate arguments etc. _______________________________________________ Boost-users mailing list Boost-users@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

The thing that I don't understand is that how folks such as those from SmartWin++ are able to seemlessly connect functions with Windows messages without using a case statement in a message loop. An example is at the following page:

http://smartwin.sourceforge.net/getting_started.php

Where do you windows messages there? All the message processing is hidden inside the framework.

Using your approach, I need to use a case statement inside a message loop.

As you can see, struct Handler in my example doesn't have any case statements as well. Again: the basic, lowlevel message processing is hidden - in my example in the "button" class, in smart win++ somewhere in WidgetWindow or its base classes. The *users* of the framework do not need to deal with messages as the framewotrk exposes higher-level facilities.

Hi Andrew, Thank you for giving me a good example. Though I haven't yet bothered to look into the SmartWin++ source code but I suspect they have done something more smarter. Thank you all. -Asif On Thu, Mar 24, 2011 at 7:12 PM, Andrew Holden < aholden@charteroaksystems.com> wrote:

On Thursday, March 24, 2011 8:42 AM, Igor R wrote:

...

...

Using your approach, I need to use a case statement inside a message loop.

As you can see, struct Handler in my example doesn't have any case statements as well. Again: the basic, lowlevel message processing is hidden - in my example in the "button" class, in smart win++ somewhere in WidgetWindow or its base classes. The *users* of the framework do not need to deal with messages as the framework exposes higher-level facilities.

Perhaps you can replace the switch statement with a map or similar (warning: untested):

typedef std::map MessageMap; MessageMap message_map;

LRESULT CALLBACK WindowProc( __in HWND hwnd, __in UINT uMsg, __in WPARAM wParam, __in LPARAM lParam ) { MessageMap::const_iterator handler = message_map.find (uMsg); if (handler != message_map.end()) { return handler->second (hwnd, uMsg, wParam, lParam); } else { return DefWindowProc (hwnd, uMsg, wParam, lParam); } }

You could then insert any handler for any message into message_map. I'm not sure what combiner would be most appropriate here. There is also the question of where message_map should really go. A global variable (as my example implies) probably isn't the best place. _______________________________________________ Boost-users mailing list Boost-users@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

Hi Igor, Thank you for replying yet again. On Thu, Mar 24, 2011 at 5:42 PM, Igor R wrote:

All the message processing is hidden inside the framework.

Basically, I want to use this library but it (and some of its examples) is not building on the Express version of Visual C++ 2010. Yesterday, I built it using the Ultimate version and it built perfectly fine without any problems. They *are* using boost and I really like their design and wish if were available on *nix as well. If I did embark on writing my own library, I'd try to do it like these guys have done it. The *users* of the framework do not need to deal with messages as the

framewotrk exposes higher-level facilities.

Yes. Thanks. -Asif