A strange question, I know.

Now I have done a fair amount of searching on this, however, I am not even sure if they have a name for what I am attempting to do... so any help in direction would be greatly appreciated, even if you simply provide the search words that might yield some actual results.

What I am trying to do is regarding the Win32 API (not using MFC), involving the message handling procedure and the UINT message parameter for the call. I am wondering if there was a way to convert the definition (for example) of WM_CREATE in to an actual string that I could change its letters, then once again use the newly created string to produce another definition's value?

Here's an example of what I am meaning:

Code:

---

LRESULT CALLBACK WinProc(HWND hwnd, UINT message, WPARAM wParam, LPARAM lParam) {
          Convert <message> to character string;
          Update string as appropriate;
          Convert string to (UINT);
          Use its value for appropriate purposes;
}

---

My problem is what is available for converting a UINT value to a string (say "WM_CREATE") that I could edit and then change back to a UINT for other purposes (I am attempting to re-map these definitions to correspond with values more appropriate to array positions that would allow my program to use appropriate processing methods depending on what message is coming in). Is this even a possibility other then creating a list to which can be referenced each time?

I don't think C++ supports converting named values to their string representation (name as in enum, define, or what ever). However, why is the convert to string step necessary? Why can't you convert or map WM_CREATE to (for example) WM_DESTROY directly?

Edit:
Unless you are willing to do something like

Code:

---

switch (message) {
    case WM_CREATE: s = "WM_CREATE"; break;
    //etc.
}

---

But that would still be strange.

These messages are defined as just that - defines! So it's impossible to convert them back to strings. You just have to create a string and do a switch-case and fill the string with the appropriate string.

I don't know exactly what the question is, but maybe you could use this to convert an indentifier to a string?

Code:

---

#include <stdio.h>
#include <string.h>
#define STRINGIFY(X) (( X ), #X )

int main()
{
        char arr[10];
        int ident;

        strcpy(arr, STRINGIFY(ident));
        puts(arr);

        return 0;
}

---

Code:

---

#include <iostream>
#include <string>
#define STRINGIFY(X) (( X ), #X )

int main()
{
        std::string arr;
        int ident;

        arr = STRINGIFY(ident);
        std::cout << arr << std::endl;

        return 0;
}

---

That macro doesn't compile, and anyway the OP is not interested in the name of the variable (message), but the name of the value it contains.

It might be best to post some more context of the question. Why do you need something like that?

Are we playing C again? We could easily use std::string if that sample would work. I'm unsure how it works.
arr = STRINGIFY(ident);

I think there is going to be no easy way to complete this project, at least in the method that I wanted to.

The purpose of the conversion from UINT to string, then back to UINT was simply for this.

Code:

---

Func* ptrArray[207]; // set to various values as required by the developer
.
.
.
WndProc() {
          char* string=ConvertDef2String (say WM_CREATE to "WM_CREATE");
          string[1]="W";
          msg=MAKEINTRESOURCE(string); // I believe this is what would do it
          if( *ptrArray[msg] != NULL ) { *ptrArray[msg](params); }
          else { DefWindowProc(params); }
}

---

But now that I have realized that any number of definitions could technically hold the same values as each other, there is no way that there could be a single method of simply identifying which string could possibly match that particular value. So I have resolved to simply creating int getArrayPos(msg), or something of the like that will reference array positions to the possible msg values and return the appropriate array position. Reference List of some kind basically. Unless someone else can think of something? I can't get the STRINGIFY example to compile on my system either.


PS - Elysia, yes I most likely am ... a hard habit to break. I'm still having issues abandoning #include <stdio.h> and all the functions that come with it.

Code:

---

WndProc() {

---

Doesn't work in C++. All functions must explicitly return something.
Anyway, what you might want is a map.

Actually, this sounds a lot like you want to do function pointers for specific messages, does it not?
Perhaps this thread might help: http://cboard.cprogramming.com/showthread.php?t=98360

In case people missed it, I fixed the macro.

You're still writing C code in the C++ forum.

Macros are the same. The C is just example code. but I'll change it for you. :)

...Except it's not good to post C code in the C++ forum and vice versa, yes? :)

It's not preferred, agreed.

Elysia,
I realize that WndProc() doesn't work in C++, it doesn't work in C either. Its a psuedo/code mix just to display what I was attempting to do, in actuality it should look like LRESULT CALLBACK WndProc(HWND hwnd, UINT msg, WPARAM wParam, LPARAM lParam), but I just placed WndProc() for easier typing. A case of laziness, nothing more.


Sorry, just realized the last portion of what you typed. Yes I am trying to create an array of function pointers that will allow me to call each repectively. I will have to look in to that, it might just be what I'm looking for. Thanks again!

I would be all for using a std::map to convert form one Windows message type ID to another, but not to and from strings.
That to me seems like an exercize in Softcoding, for which you'd need to read this:
http://thedailywtf.com/Articles/Soft_Coding.aspx
http://en.wikipedia.org/wiki/Softcoding

Yeah, I proposed maps and made an example of a working system in the thread I linked to.

Actually since I am seeking technically a map between one definition to another (both of which are number based, even though they technically use different mathematical bases) I think a map would do just fine for what I need to accomplish.

0x034 (e.g.) needs to be mapped to 15

thus when the msg 0x034 comes in to my message handler, I access the appropriate function pointer via... ptrArray[15](parameters);

Thanks for the advise Elysia, a great help as always :)

Well, you don't really need to re-map the number of message at all. Since all messages have unique numbers, you can just map them to a function pointer directly, if you wish and pass any argument you think may be useful.
The map example I made in the other thread is just a basic map-to-pointer example. If you modify the function pointer typedef, you should be able to map whatever number you want into appropriate function pointer and function handler.
It's open to modifications.

Perfect. I guess I was so wrapped up in my own concept idea, it didn't even occur to me to take my own defines out of the equation. Thanks again :)

Quote:

---

Originally Posted by cyreon

Perfect. I guess I was so wrapped up in my own concept idea, it didn't even occur to me to take my own defines out of the equation. Thanks again :)

---

It is quite common to "get stuck in the idea". Sometimes it helps to explain a bigger picture to someone else, and the other person may come up with a better idea of how this can be done [which is why I often ask for "what's the big picture" in threads where someone is asking "How do I ..."].

Generally discussing ideas for solutions help, of course.

--
Mats

It's wonderful to explain the bigger picture to someone.

"So I need to make the ball hop down the hall and therefore I need to ... need to ... what am I doing this nonsense for?"

Quote:

---

Originally Posted by CornedBee

It's wonderful to explain the bigger picture to someone.

"So I need to make the ball hop down the hall and therefore I need to ... need to ... what am I doing this nonsense for?"

---

Oh, yes, the number of times I've had that experience: Explain something clearly to someone else, and you immediately realize what you are doing wrong...

--
Mats

The nice thing that it works even with non-programmers. I habitually explain my problems to my father, who can barely follow me. Still helps.