Thread: Ask me about C / C++

How do I write a callback function that will get called at a rate that I can define?

Originally Posted by khh

Yeah. That's not at all how I imagined they would work, but it seems fairly logical now that you present the idea. (Though I had to look spend some time just looking the code over and mentally executing it to get the picture :p)

How would you include user defined functions? One way of doing it could be to just paste the code for each time it's used (like inline functions), but how would you get proper functions?
Also, I'm guessing it would be pretty essential to implement a syntax checker?

I actually don't implement a syntax checker, my company has a dev-staging-live process, my app is live so it's assumed that everything is correct when it gets to me.

I do implement inline functions in that very way, and as for user defined real functions, well, my current language doesn't support those at this time, but I plan to add them in the very near future. They aren't that difficult. Basically, you would just replace a call to that function with something like @UserDefined(<place>) during a preprocessing pass. Then when you compile your callbacks, you have it point to the function that you've defined.

Yeah, scripting languages are weird things, and difficult to start implementing, but once you've done a few of them, you can write pretty advanced ones really quickly.

Originally Posted by Artelis

How do I write a callback function that will get called at a rate that I can define?

You mean in a scripting language, or in C++?

Originally Posted by ninja9578

I actually don't implement a syntax checker, my company has a dev-staging-live process, my app is live so it's assumed that everything is correct when it gets to me.

I do implement inline functions in that very way, and as for user defined real functions, well, my current language doesn't support those at this time, but I plan to add them in the very near future. They aren't that difficult. Basically, you would just replace a call to that function with something like @UserDefined(<place>) during a preprocessing pass. Then when you compile your callbacks, you have it point to the function that you've defined.

Yeah, scripting languages are weird things, and difficult to start implementing, but once you've done a few of them, you can write pretty advanced ones really quickly.


You mean in a scripting language, or in C++?

in a scripting language...

Isn't this a C/C++ thread?! Why would I be asking questions that AREN'T about C or C++? Well anyway, I'd prefer it in C, but I'm pretty sure it'd be handled the same way in either. Callbacks don't really use classes from my knowledge.

Callbacks are just regular methods. All a method is is a pointer to a bit of code, nothing more, so you can store a pointer to a method just like to store a pointer to an object.

Code:

someclass * a = new someclass();
someclass * b = a; // referencing a class

Code:

typedef int (*addcallback)(int, int);  //declare the type of callback and the parameters

int ADD(int a, int b){  //the actual method which you will be calling back to.
  return a + b;
}

//in anywhere else in the same namespace
addcallback add = ADD;  //make a reference to the method, just like you reference a class
int j = add(4, 5);  //use the callback just like you would any other method

What did you mean "that gets called at a rate that I define?" You can call the callback whenever you'd like.

function pointers are a good thing to get your head around for advanced stuff.
Function pointers are at the heart of both polymorphism and function overloading
you can do some seriously cool stuff with function pointers

Originally Posted by Ynot

function pointers are a good thing to get your head around for advanced stuff.
Function pointers are at the heart of both polymorphism and function overloading
you can do some seriously cool stuff with function pointers

Also shared libraries. Modularization is one of the most important things in advanced programming. No big program is one big program, it's a little program and a multitude of helper dlls or sos.

I'm looking for a cross-platform library implementing network connections, and a good guide on how to use it. If boost has such networking functionality that's be good.
Basically what I want is a library which already implements the whole TCP/UDP. I'm just looking to create some communication between a generic client and a generic server. Know where I can find that?

Yes, Boost has a network library. Boost Network I've never used it, but I know that it exists and because it's Boost there are lots of tutorials for it.

also have a look at QT
it's much more than a GUI layer
(in fact, you don't even have to use the GUI layer - you can just use the networking libraries)

wxWidgets too. I prefer wx because it's got a huge community for support, whereas Qt used to be a commercial product, so support is limited.

I changed my mind about wxWidgets. Don't use it, it's given me my last headache, I'm writing my own library from scratch, if anyone wants to help, let me know.

Okai, I have a question... again :p

So in a recent school assignment I need to dynamically allocate a 2D table. I chose to do it like this:

Code:

int** table;
table = new int*[width];
for (unsigned n = 0 ; n < width ; ++n) {
	table[n] = new int[height];
}

(if there is a better way to do it, please tell me :p)

But I had a problem when I was deallocating the memory again. I had thought I could simply do it like this:

Code:

for (unsigned n = 0 ; n < width ; ++n) {
	delete[] table[n];
}
delete[] table;

but that frequently (though not allways) caused the program to crash, citing heap corruption as the cause. So why would the above crash the program, and how can I deallocate the memory?

(I realise the problem could be circumvented by allocating it as a 1D table, and write a function that can translate two coordinates to an index, but I want to know why the above doesn't work)

Well, there is a better way to do it, it's always best to allocate memory that goes together all at once. Remember, using a C-cast you can change anything into anything else.

Code:

int** array = (int**)malloc(width * height * sizeof(int));

Then to delete it simply:

Code:

free(array);

Looking at your code at a quick glance, I don't see anything incorrect. I would run it through the debugger and put a breakpoint on each delete[]

Also, it's good practice to explicity declare what type you're using.

Code:

unsigned n = 0; //not standard C++
unsigned int n = 0;  //correct

Wow, quick reply

Originally Posted by ninja9578

Well, there is a better way to do it, it's always best to allocate memory that goes together all at once. Remember, using a C-cast you can change anything into anything else.

Code:

int** array = (int**)malloc(width * height * sizeof(int));

Then to delete it simply:

Code:

free(array);

I have heard that it is good practise not to use the C functions in a C++ program, though I might have misinterpreted this. Would it be possible to use something like

Code:

int** array = static_cast<int**>(new int[width*height]);
delete[] array;

Also, how would that code know that I wanted a table of the dimentions width x height, and not height x width or something else entirely? I would want it to function identically to a static array that could be created with

Code:

int array[width][height];

if width and height were constants.

Originally Posted by ninja9578

Also, it's good practice to explicity declare what type you're using.

Code:

unsigned n = 0; //not standard C++
unsigned int n = 0;  //correct

Yeah, usually I define shorts like UINT in a header file. I did think that "unsigned" was standard C++ for "unsigned int", though, like "long" is standard for "long int". :p
But since you have pointed out it isn't, I'll try and use fully qualified variable names from now on :p

Originally Posted by ninja9578

Looking at your code at a quick glance, I don't see anything incorrect. I would run it through the debugger and put a breakpoint on each delete[]

I just added a cout statement, and found that "delete[] table[0];" worked, and then it crashed on "delete[] table[1];" I tried having n count down from width-1 instead, but the result was the same, it crashed on the second deallocation.

No, using C functions is very common. C functions are very very fast. Don't use C headers. ie #include <ctime> instead of #include <time.h>

Anyway, if I were you, I would write a structure to handle an array for you and a method to get specific items:

Code:

#include <iostream>

struct Array {
public:
    Array(unsigned int width, unsigned int height){
	   w = width; h = height;
	   array = (int*)malloc(width * height * sizeof(int));
    }
    ~Array(void){
	   free(array);
    }
    inline int & at(unsigned int x, unsigned int y){
	   assert(x < w);
	   assert(y < h);
	   return array[w * y + x];
    }
private:
    int * array;
    unsigned int w, h;
};

int main(){
    Array test(100, 150);
    test.at(12, 43) = 15;
    std::cout << test.at(12, 43) << std::endl;
    return 0;
}

Oops. I forgot a const at operator

Code:

    inline const int & at(unsigned int x, unsigned int y) const {
	   assert(x < w);
	   assert(y < h);
	   return array[w * y + x];
    }

Did you get how everything in my code works? This is how arrays are done behind the scenes anyway and because it's just a wrapper and the only functions are inline, the compiler will probably optimize away the structure.

Mostly, yeah. But I don't get why you use assert. I thought the point of an assert statement was to halt program execution. Would it not be more suitable to use an if statement?

Also, what does the const operators do in the last bit you posted? I have never learned to use them, except in variable declarations.

Yes, asserts halt operation, but only in debugging mode. In release mode, they are compiled out, so no checks are done. This helps you make your code correct before you release it.

const at the end of a function means that object that it's being called on will not change. When variables are declared as const, it means that they can't be changed. So you can only call const functions.

Code:

#include <iostream>

class foo {
public:
    foo(void){};
    ~foo(void){};
    void Test(void) const {
	   //not allowed to change anything in this class
	   std::cout << "Test const" << std::endl;
	   //bar = 1;  <- This is illegal
    }
    void Test(void){
	   //can manipulate the object
	   std::cout << "Test" << std::endl;
	   bar = 1; //This is fine
    }
private:
    int bar;
};

int main(){
    foo one;
    one.Test();  //foo is not const, so the non-const method will be called
    const foo two;
    two.Test();  //foo is const, so the const method is called
    
    return 0;
}

const essentially means "read only" so
operator [] lets you read and write to the class
operator [] const lets you read from the class only

Okai, I understand. Thanks

I thought I should pop up a slightly old thread for this problem I'm having.

I started *trying* to use wxWidgets.
IT makes sense... a little bit.

But there is one roadblock in my way of progressing in a program I am trying to get done.

Basically, I want to be able to click a button (In this case, "Button2"), so that the text changes. I can do that fine. But I then want to be able to click it again so that it changes to something new.


I have tried countless different possible ways of what I thought would work, but I just can't get it.

This was my last attempt (The bit I'm at a problem with):

Code:

void Physics_StudyFrame::OnButton2Click(wxCommandEvent& event)
{
    for (int a=0; a<2; a++)
    {
        (StaticText2)->SetLabel(_("FirstClick new Text"));
        a = a + 1;
    }

}


void Physics_StudyFrame::OnButton2Click2(wxCommandEvent& event)
{
    (StaticText2)->SetLabel(_("Second Click New Text"));
}

I dunno, In my head it seems that should work. But it just doesn't.

I thought about possibly making the button disappear on first click and then a new one re-appear, so that for the second click I could just use the new button. But... I don't even know where to begin for that one

Halp

Originally Posted by slash112

Code:

void Physics_StudyFrame::OnButton2Click(wxCommandEvent& event)
{
    for (int a=0; a<2; a++)
    {
        (StaticText2)->SetLabel(_("FirstClick new Text"));
        a = a + 1;
    }

}

I didn't look at any of your code, but this popped up immediately: why are you incrementing a twice? Once as part of the for loop (a++), and once at the end (a = a+1). while I have no idea what you're trying to do, I'd be shocked if this weren't seriously broken.

My guess is that you are using an event table for this, I would recommend connecting your events and disconnecting them manually.

When you create it, do a ConnectEvent, on the first click, do a DisconnectEvent, then ConnectEvent to have it go to the new method.

Originally Posted by Replicon

I didn't look at any of your code, but this popped up immediately: why are you incrementing a twice? Once as part of the for loop (a++), and once at the end (a = a+1). while I have no idea what you're trying to do, I'd be shocked if this weren't seriously broken.

A series of stupified attempts brought me to this equally stupid one.
Basically, I thought that I could use a loop, to make it sort of stop... or something. I dunno, it seems more stupid when I start thinking about it more...

Originally Posted by ninja9578

My guess is that you are using an event table for this, I would recommend connecting your events and disconnecting them manually.

When you create it, do a ConnectEvent, on the first click, do a DisconnectEvent, then ConnectEvent to have it go to the new method.

Ahh, well this makes sense, actually.
*Goes to Google*