|
Application Wide Central Hotkey Filter based on Win 32 API Hooks
I have written a central hotkey manger prototype (see describtion below the question) for a complex MDI Application (child windows / subwindows are loaded via COM and other technologies). A simple Win 32 API hook (e.g. WH_KEYBOARD_LL) based filter (written in C# but that does not matter - could also be C/C++) replaces a lot of keyboard event forwarding code in an applications sub windows written in many different languages (as they are plug-ins etc.) and you no longer have to deal with different keyboard event management concepts in the different UI Frameworks used.
But I have two problems (the cases A and B below) as I am not a Win32 API Expert.
----------
QUESTION
I would be grateful for any ideas about how the states/cases below can be determined !easily! via Win32 API, without having to change the code of the sub windows.
A) "covered by a modal sub window"
I have to know (without changing the code of the modal dialog) when in the context of the main window (to that hotkeys are forwarded) a modal child window / dialog is opened (and closed), because the hotkey filter shall not filter any keyboard events (event when hotkeys) in the context of modal dialogs.
B) active parent window
You can think of cases, when there are several “main windows” ("main" means: windows that receive hotkeys and can have child windows)
that receive application wide hotkeys (but each main window belongs to a different sub tree in the window hierarchy), all main windows distinguished by their win32 API handles. The handles of the main windows are known to the hotkey manager, but not the sub windows.
Main Window A --child--> SubWindow2 … --child--> SubWindowFocus
Main Window B --child--> SubWindow3
When “SubWindowFocus” has the Focus each keyboard event reaches the same Hotkey Manager via the keyboard hook, the hotkey manager only knows (handles) A and B.
Is it possible to determine the appropriate Main Window (A would be correct) via Win32 API? In that context the event was raised.
-----
PROTOTYPE – Application Hotkey Filter
I am actually developing a “hotkey manager” that works in a complex environment (i.e. the main system and main windows are written in smalltalk or c++ and there are lots sub windows (i.e. plug-ins and add-ons) implemented in Java, .NET, VisualBasic …
Regarding the forwarding of not consumed keyboard events from the sub windows to the main windows there are many bugs and problems (some framework related or COM related) in order to get rid of all these problems I have written a central hotkey manager (filter) that attaches itself to each relevant event dispatching thread of the application.
The manager uses a Win 32 API keyboard hook (a local hook) that filters all keyboard events before the reach the sub windows, the filter than redirects hotkeys directly to the main window instead of relaying on the sub windows to forward them. Non-hotkeys go the normal way.
My code is actually quite ugly (prototype) so it makes no sense to post it (i will write a hotkey filter article when finished) - if you are interested in this topic see the codeproject article (this a global hook, i used a local one) on that my solution idea is based:
http://www.codeproject.com/KB/system/CSLLKeyboard.aspx
Best regards,
Martin Schwienbacher
modified on Tuesday, January 15, 2008 4:59:15 AM
|
|
|
|
|
Hello all,
I am pretty much excited to get registered to this site because its totally different from other programming sites...I saw codes and demos of some windows application.It was mind blowing to see the UI and the code usage...
I am new to c#...In the sense,i have developed some simple applications like user authenticatoin,insertion,updation,deletion,search,etc...
Those things can be done in any language...I really like working with c# and want to dive in deep...So far i know about the basic controls and simple codes....I need help from u peopele like what are the basic things and how to develop application like a professional..I have some books on c# but those books just brushes the basics of all the controls..I want to develop my programming skills and good practices on how to code and design.
So suggest me some pretty good books on developing Gui application in c#..
Thanks and Regards
Bharathi..
|
|
|
|
|
You can start with "C# Step by Step " by Microsoft Press.
Good luck !!!
|
|
|
|
|
One suggestion is to study Design Patterns in C#. It may take you a year to get up to the level of completely understanding everything the book presents, but you will keep it by your side the entire time you use C#.
|
|
|
|
|
How to remotely connect multiple PCs(Windows OS based) computers from a single Mac and display all of them on Mac Leopard Spaces !!
can somebody give me a good suggestion on this,
Thanks You ,
Nickle
Happy Programming!
Regards,
Nikhil Pagidala
|
|
|
|
|
Nikhil,
You kind of went backwards in your approach to getting information. Instead of coming to a C#, Windows development forum, you should go to a MAC forum and ask there. Your code has to be written and consumed within the MAC.
|
|
|
|
|
I have a grid control with a number of properties, including colCount, rowCount and rowFooterFillType.
When I change them in designer, they get written into <class>.designer.cs in alphabetic order.
The initialization happens in this order:
newGrid.colCount = 8;<br />
newGrid.rowCount = 4;<br />
newGrid.rowFooterFillType = Type.typeOf("System.Windows.Forms.Label");
What happens is that the grid is created, it is sized, and then it finds out that each of the columns was supposed to create a footer of type "Label".
It makes more sense for the code to be initialized like so:
newGrid.rowFooterFillType = Type.typeOf("System.Windows.Forms.Label");<br />
newGrid.colCount = 8;<br />
newGrid.rowCount = 4;
That way, when the rows are created, they already know what type of footer to use, and create them like I want.
I can go in and change the order of the code designer put out, but that does no good the next time I change one of the attributes in designer, since designer puts them back into alpha order.
Anyone got any good ideas?
Thanks
I want to die like my Grandfather.
Peaceful, Sleeping.
Not screaming like his passengers.
|
|
|
|
|
I suggest you don't attempt to control the order the properties are called in. In fact, this is a violation of the FxCop (.NET rules) specification - properties are supposed to be black box so that it doesn't MATTER what order you set them in.
That said, it is obviously unrealistic to expect properties to not affect each other. This is why the designer supports the BeginInit() and EndInit() methods from the ISupportInitialize interface. The designer will automatically insert BeginInit() before it sets any of your properties, and will call EndInit() when they're done.
What you can do with this is when BeginInit is set, set a flag in your code to true, and as long as it is true, simply update your backing field for the properties. Don't do any processing on them.
When EndInit() is called, unset your flag, and execute all of the code that should be executed to make all of your properties take effect.
All you need to do to support this behavior is add " : ISupportInitialize" to your class declaration and implement the appropriate methods.
Read more about it here[^]
It has become appallingly obvious that our technology has exceeded our humanity. - Albert Einstein
|
|
|
|
|
THANK YOU!
I know it is not the best behavior, but I am setting default types for the grid cells, and I sort of need to know what they are BEFORE I create the cells, since the user might change the default for new cells, without wanting to re-initialize the old cells.
ISupportInitialize looks like it will be just the trick.
I want to die like my Grandfather.
Peaceful, Sleeping.
Not screaming like his passengers.
|
|
|
|
|
RichardM1 wrote: I know it is not the best behavior, but I am setting default types for the grid cells, and I sort of need to know what they are BEFORE I create the cells, since the user might change the default for new cells, without wanting to re-initialize the old cells.
Nah, it sounds fine. Sometimes this is the simplest approach, even if it seems like it's not elegant enough. I've learned that if it takes 32 hours to come up with an elegant solution, when a less-elegant-but-equally-workable-solution will do, implementing the elegant solution is a waste of time.
ISupportInitialize is pretty cool  I don't use it often, but when I do need it I'm damn glad it's there..
It has become appallingly obvious that our technology has exceeded our humanity. - Albert Einstein
|
|
|
|
|
LOL!
It is like a hammer:
Sometimes a jeweler's screw driver just will not make the TV work, so you have to use the hammer.
I want to die like my Grandfather.
Peaceful, Sleeping.
Not screaming like his passengers.
|
|
|
|
|
I need help with following problem:
I created new project with two Windows Forms called Form1 and Form2. Now, I put on each Form one Button and one Text Box.
The main Form is Form1 which I start the program. When I click Button1, Form2 shows and into Textbox1 in Form2 it puts some text. To do this in Form1 I added code:
#include "Form2.h"
...
private: Form2^ form2;
...
private: System::Void button1_Click(System::Object^ sender, System::EventArgs^ e)
{
form2 = gcnew Form2();
form2->textBox1->Text="sometext";
form2->Show();
}
Everything works fine, but..... Now I want do the same in opposite direction. I want to click on Button1 on Form2 and put some text into textbox1 on Form1. How to do this????
I cannot add in Form2 #include "Form1.h" because it will not compile(There are some errors)
Please help me with this "simple" problem.
Best regards
Szymon
|
|
|
|
|
First, you really need to break the habit of a Form directly modifying another Form. Comming from a C++ world, you should already know about OOP and encapsulation. What you're doing is breaking the rules of encapsulation for sure because a Form (class actually) needs to know about the internals of another class. You're explicitly tying those two classes together to the point where one cannot exist without the other. Does that sound like good encapsulation to you??
Get into the habit of creating a data model first, completely independent of a user interface. I know it's a bit overkill for such a simple example as what you're doing now, but the experience comes in very handy when your projects become larger and larger, and when you need to run the same app as both a Windows Forms app and as a Web Forms app.
For your case, a class that exposes a property that hold the "message" being passed between forms would work just fine. This property would fire an event that either (or both) forms can subscribe to so that if either form, at any time, set the property (call it "Message"), both forms (or even more) would get notified of the change and be able to update their own UI components themselves. Neither form has to know anything about the other.
|
|
|
|
|
Thanks, Could you put here example of code, which I can use? How would my code looks then to obtain task I want?
Szymon
|
|
|
|
|
A technique is to have an object that holds the data to be transfered.
For instance, a class "person", with "name", "birth date", and other fields.
For certain of the properties of person, like "birth date", you want change events.
Form 1 creates a person, and registers with the person's change events.
Form 1 then creates form 2 and passes it the person.
Form 2 knows what it wants to know about the person, and uses it, and may register itself with some of person's change events.
If form 2 has a button "Age Person" that makes person's birthday a year earlier, that will fire the change (in person), which will notify form1 that it needs to update.
Form 2 knows nothing of form 1.
Form 1 knows nothing of form 2.
Both only know what they need to about Person, and nobody knows anyones else's internals.
Person doesn't know about either of them
I want to die like my Grandfather.
Peaceful, Sleeping.
Not screaming like his passengers.
|
|
|
|
|
OK, I understand. So I don't need to include "Form.h" to have access to Controls of Another Form?
So if Form1 has Form1.h and Form2 has Form2.h and I declare a class let say "person" inside Form1.h. Then I create Form2 and in Form2 I change properties of person, I must declare somewhere in Form2.h that "person" class exist. In other case how Form2 will notify Form1 that properties of "person" were changed. How do I declare 'existence' of "person" in Form2??? Are you able to put some code?
Thank you
Szymon
|
|
|
|
|
You have a seperate "person.h" that defines the class "person".
Both form1 and form2 include person.h.
Form1 also includes form2.h, since it will have to create one, but it never needs to know more than the constructor.
Form2 updates itself from person, and from change events from person. Form1 does the same. So again, they have no knowledge of the internals of the other.
Bear with me on this, I have not done c++ in a while, and not without an IDE in an even longer time, so this is more psuedo code than code, or maybe it is a hybrid with c#.
person.h
class person<br />
{<br />
public person(string name, DateTime birthday);
void SetName(string name);<br />
string GetName();<br />
void SetBDay(DateTime birthday);<br />
string GetBDay();<br />
void AddBDayChangeCallBack(callBackType callBack);<br />
void DelBDayChangeCallBack(callBackType callBack);<br />
}
person.cpp
class person<br />
{<br />
private string m_name = "";<br />
private DateTime m_birthDay = new DateTime();<br />
private callBackTypeList m_bDayChangeCallbacks = new callBackTypeList();<br />
<br />
public constructor person(string name, DateTime birthday)<br />
{<br />
SetName(name);<br />
SetBDay(birthday);<br />
}<br />
void SetName(string name){}<br />
string GetName(){}<br />
void SetBDay(DateTime birthday)<br />
{<br />
m_birthDay = birthday;<br />
foreach(callBackType event in m_bDayChangeCallbacks)<br />
{<br />
invoke(event(this)); <br />
}<br />
}<br />
string GetBDay(){}<br />
void AddBDayChangeCallBack(callBackType callBack)<br />
{<br />
bDayChangeCallbacks.Add(callBack);<br />
}<br />
void DelBDayChangeCallBack(callBackType callBack)<br />
{<br />
bDayChangeCallbacks.Remove(callBack);<br />
}<br />
}
So form1 and form2 both include person.h.
form1.h and form2.h do not expose any of the components of the forms, just constructors/destructors.
The form2 constructor takes a person as a parameter, and that is how it is notified of the person, though you could also have a "void SetPerson(person perp);" definition somewhere.
Wow. That is too bad they flatten indentation.
I want to die like my Grandfather.
Peaceful, Sleeping.
Not screaming like his passengers.
|
|
|
|
|
Person exists in its own right, it has no relationship to any form, hence its class
is not inside Form1, it is in a separate file, say Person.h
Form1 and Form2 are ways to interact with Person, they are separate classes in separate
files.
Doing so, both Form1 and Form2 need to know about Person, but not about each other.
|
|
|
|
|
MUCH more condensed and readable than what I said.
I want to die like my Grandfather.
Peaceful, Sleeping.
Not screaming like his passengers.
|
|
|
|
|
That is because Luc is the MAN. (congrats on the MVP by the way, Luc!)
It has become appallingly obvious that our technology has exceeded our humanity. - Albert Einstein
|
|
|
|
|
Thanks. I try and help where I can.
|
|
|
|
|
And no code at all, since I don't speak C++ fluently... 
|
|
|
|
|
(shaking head)
If you look at what I wrote, it's probably obvious I don't, as well.
I want to die like my Grandfather.
Peaceful, Sleeping.
Not screaming like his passengers.
|
|
|
|
|
RichardM1 wrote: it's probably obvious I don't, as well
I truely can't tell, at least until I pass it to the compiler.
I did not really study your post, we both were replying concurrently.
BTW: you can (and should) preserve formatting by using PRE tags
(CODE tags are good for single-line captions, PRE is useful for multi-line).
|
|
|
|
|
Thank you Sir, you are a scholar and a gentleman.
I want to die like my Grandfather.
Peaceful, Sleeping.
Not screaming like his passengers.
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
