C / C++ / MFC

Re: Help me please! Release mode optimizations are screwing up my program! :(

16-Feb-02 19:39

Life is Full of Suprises
VC++ in debug mode assignes non-NULL values to newly allocated memory as a debug helper. This explains why it works in debug.

pszBuffer should not be NULL. I'm not sure why i told you to get rid of the line that checked for that. (not sure why i'm still on CP at 12:30AM either... great, another addiction)

However, the data pointed to by pszBuffer might easily be NULL. Especially the first character, since this would be the NULL terminator on an emtpy string. Internally, CStrings maintain a length counter, so the actual databuffer can be any size. The NULL terminator is strictly for the convenience of C-style string handling routines that need to access the CString. Keep in mind how the standard C library functions work: strcpy() does not require a length specifier, but is a danger spot for buffer overruns if you are not careful, since it does not (and cannot) check that the destination buffer is large enough to hold the source string. strncpy() does require a length specifier, and uses that to operate safely.

This is why using a string class instead of raw C-style strings is nice; constantly having to make your code aware of the total size of buffers gets old in a hurry.

Anyway, sorry you have to be working on the weekend; good luck on your project!

farewell goodnight last one out turn out the lights

Smashing Pumpkins, Tales of a Scorched Earth

Julia Larson16-Feb-02 20:12

Julia Larson

16-Feb-02 20:12

Thanks, you have totally saved my life. I was full-on panicking
earlier this evening. OMG | :OMG:

whew!

It slowly dawns on me

Julia Larson16-Feb-02 19:16

Julia Larson

16-Feb-02 19:16

I think I get it.

The content of the pointer could be anything, as you said.
Even though I said I wanted to allocate a certain amount
for a buffer using GetBuffer, the contents of the buffer
could be anything - including '\0', but it wouldn't be a '\0'
that I had asked for or put in there. It would just be random
memory. Oh jeez. weird. ok.

Thanks again,
-Julie

CDialogBar

16-Feb-02 16:28

Hello,

I have a subclass from CDialogBar. I try tu put a IP control in it, but i can´t get the bytes in that control.

I see that happens not only with this control, it happens with any control that don´t be a edit...

What happens?

Thanks Codeprojects...

UI

Re: CDialogBar

Derek Waters17-Feb-02 11:27

Derek Waters

17-Feb-02 11:27

Can you post some code? I've used CDialogBars quite a bit, and have never experienced the problems you are talking about.

------------------------
Derek Waters
derek@lj-oz.com

Drawing Controls Problem

John Clump16-Feb-02 16:13

John Clump

16-Feb-02 16:13

Hi everyone. I have some code which centers a series of buttons on a dialog, which is in OnSize(). Problem I have is, as the user resizes the control, the buttons are drawn twice (?), and there are "artifacts" of the original draw are still there. If the window is minimized and brought back up again, the buttons draw properly and the "artifacts" are gone. The buttons do center and adjust properly as the window is resized, but leave those "artifacts" for some reason. It doesn't happen when the window is maximized and normalized, but do occur when the user drags the window frame and resizes that way.

I am also using Paul Wendt's CControlPos class which is found on Codeproject, which moves all the other controls (I had to move the buttons manually to achieve the centering with a series of buttons).

Anyone have any ideas? Any help is greatly appreciated!

<br />
void CDlgMain::OnSize(UINT nType, int cx, int cy) <br />
{<br />
	CDialog::OnSize(nType, cx, cy);<br />
<br />
	CRect rcButtonRect;<br />
<br />
	// size once window is created, then move buttons appropriately<br />
	if(::IsWindow(GetDlgItem(IDC_BTN_ADD)->GetSafeHwnd()))<br />
	{<br />
           // resizing calculations done here<br />
<br />
		bool bDrawAgain = true;<br />
<br />
		// draw add button<br />
		GetDlgItem(IDC_BTN_ADD)->MoveWindow(rcButtonRect, bDrawAgain);<br />
<br />
		// draw edit button<br />
		rcButtonRect.left = ((cx - iParentWidth)/2) + iButtonLeft;<br />
		rcButtonRect.right = rcButtonRect.left + iButtonWidth;<br />
		GetDlgItem(IDC_BTN_EDIT)->MoveWindow(rcButtonRect, bDrawAgain);<br />
<br />
		// draw delete button<br />
	        rcButtonRect.left = ((cx - iParentWidth)/2) + iButtonLeft;<br />
		rcButtonRect.right = rcButtonRect.left + iButtonWidth;<br />
		GetDlgItem(IDC_BTN_DELETE)->MoveWindow(rcButtonRect, bDrawAgain);<br />
<br />
		// draw copy button<br />
		iButtonLeft = 391;<br />
		rcButtonRect.left = ((cx - iParentWidth)/2) + iButtonLeft;<br />
		rcButtonRect.right = rcButtonRect.left + iButtonWidth;<br />
		GetDlgItem(IDC_BTN_COPY)->MoveWindow(rcButtonRect, bDrawAgain);<br />
	}<br />
    // Paul Wendt's CControlPos class, this moves all other<br />
    // controls<br />
    // this class calls the following code to draw each<br />
    // control<br />
    // m_pParent->ScreenToClient(rcBounds);<br />
    // pControl->MoveWindow(rcBounds);<br />
<br />
    m_cControlPos.MoveControls();<br />
}<br />

Shog⁹16-Feb-02 16:25

16-Feb-02 16:25

Is WS_CLIPCHILDREN set for the dialog?
Maybe calling CDialog::OnSize(nType, cx, cy); after moving the controls would work.

farewell goodnight last one out turn out the lights

Smashing Pumpkins, Tales of a Scorched Earth

John Clump16-Feb-02 18:00

John Clump

16-Feb-02 18:00

I tried doing both, with no luck Frown | :(

It still leaves junk behind when the user drags on the window frame

Shog⁹16-Feb-02 18:30

16-Feb-02 18:30

well, you could also pass in FALSE to all the MoveWindow() calls, then follow everything with Invalidate(TRUE). It'll flicker a bit, but should look ok.

farewell goodnight last one out turn out the lights

Smashing Pumpkins, Tales of a Scorched Earth

Ravi Bhavnani16-Feb-02 19:19

Ravi Bhavnani

16-Feb-02 19:19

I've also seen this problem when resizing static picture controls (eg: frame borders). Try invalidating and updating the dialog at the end of the repositioning.

void CDlgMain::OnSize(UINT nType, int cx, int cy) 
{
  CDialog::OnSize(nType, cx, cy);
  ....  // rest of your code
  Invalidate();
  UpdateWindow();
}

/ravi

"There is always one more bug..."
http://www.ravib.com
ravib@ravib.com

Help on Designing Tictactoe in VC++

John Clump16-Feb-02 19:47

John Clump

16-Feb-02 19:47

That worked great! Thanks for the help!

John Cruz16-Feb-02 15:38

John Cruz

16-Feb-02 15:38

i am trying to write a tictactoe game in VC++ but i dont know how to organize it and whats the best way to do it. i am planing to use Bitmap graphics and beneath it use a grid to find out which one the user clicks. and i am planing to use recursion to implement it but i dont know if thats the efficient way to do it. can anyone help me. i am not good in designing? can someone give me an advice on whats the best way to do it?

Shog⁹16-Feb-02 16:18

16-Feb-02 16:18

This is your homework, isn't it. Suspicious | :suss:

Here's what i'd do to get started:
use the VS wizards to create a dialog based app. Delete all controls from the dialog. Implement an OnPaint() handler that draws your board. Handle mouse messages to respond to player actions. Use a separate class to implement game logic.

There are many ways to implement a tic-tak-toe game; just use whatever makes the most sense to you. If you have a specific question, i'm sure you'll get plenty of answers to it.

Have Fun!

farewell goodnight last one out turn out the lights

Smashing Pumpkins, Tales of a Scorched Earth

alex.barylski16-Feb-02 22:28

alex.barylski

16-Feb-02 22:28

I did this for my Gr. 12 comp sci course last year, but in VB... Poke tongue | ;-P

It is your homework...

In anycase. This is what I did.

1) Create a dialog
2) Create (an array or not) of 9 Pushbutton controls and set the font accordingly, I used Arial Bold 23 Pts. This will let you focus on game logic, instead of x and o drawing.
3) Last create an array of 9 chars, NULL=Empty, X is X, Y is Y. This array will map to your physical buttons. this way you can disable the button once it was clicked and avoid having to double check the array to ensure someone hasn't chosen the same location twice.
4) On each buttonclick set the coresponding array element to X or Y depending on whos turn it is, then change the current buttons caption to the current elements value.
5) Design a killer AI, that of which will rival a Russian chess champion.

Tips: It'll be much easier to create the buttons statically and position them with dialog editor.

Tips: Each button click do the following!

1) Iterate array looking for X or O's horizontally
2) Iterate array looking for X or O's vertically
3) Iterate array looking for X or O's diagonally

If your test function returns TRUE you know someone has won the game!

BOOL IsWin()...???

your choice...from here on your on yer own...

Cheers!

"An expert is someone who has made all the mistakes in his or her field" - Niels Bohr

John Cruz18-Feb-02 12:43

John Cruz

18-Feb-02 12:43

Yah, i made my design using MFC. i used the CDC function moveto and line to draw the board. beneath it i made an array of Crect so i would know which one i clicked. if i click a point inside a rectangle an X will be draw using the textout function. pretty much, i got all the designing done, my only problem now is creating the AI for my tictactoe. i am thinking of using recursion to implement that. i dont know if that is efficient or just make a list of moves instead. is using recursion better? honestly this is not my homework. in school where learning assembly but i thought id practice MFC and C++.

Christian Graus18-Feb-02 12:47

Christian Graus

18-Feb-02 12:47

You're learning assembly in school ? Cool.

The most interesting and effective way I can think of would be to program the computer to play itself and remember moves that lead to a win, as well as obviously searching for the ability to finish a row. That way it will teach itself.

Sorry, I really have only vague ideas about how to do this.

Christian

I have come to clean zee pooollll. - Michael Martin Dec 30, 2001

Picture the daffodil. And while you do that, I'll be over here going through your stuff.

Andrew Peace18-Feb-02 13:09

Andrew Peace

18-Feb-02 13:09

A tic-tac-toe AI alogorithm is quite similar to a Chess AI except much simpler. There are numerous ways to implement it, and recursion is certainly an escellent bet IMHO.

Work around a frameowrk similar to this;

For each player move:
1. Determine possible computer moves
2. For each computer move, guess the player move. Let the computer mentally play the game until a given recursion limit (e.g. 6 moves) is reached.
3. For each series, return a score for that path of play.
4. Collate the data, and the highest scoring move-set wins, and the move is made.

Optimisations can be made, such as after the move tree has been compiled, if the player makes the predicted move, just chop off the inappropriate branches and iterate another level. This is quite a complex optimisation if you are only starting out, but will probably be worthwhile if you want to go that far.

Remember, if the board is empty, and you are playing on a 3x3 grid, there are 3x3 possible moves, i.e. 9 moves. However, once the computer has moved (presuming the computer were to take first play), there will be eight remaining moves for the player. Bearing in mind that the computer will have to guess the player's moves as well as it's own, that means that, if you take to recusion to 6 levels, there are 9P6 possible moves, (i.e. 9x8x7x6x5x4 moves). For a full game, there would be 9! possible moves, i.e. 362,880 possible moves, so you probably wouldn't want to guess out the entire game.

OK - moving away from the actual calculations that have to be done, how would you store this data. If you were to use a tree structure, each node in the tree would need the move data (i.e. X and Y co-ords), a pointer to the next sibling node, a pointer to it's first child node and possible a parent-node pointer. Hence, you would need something like;

class CMoveNode {
public:
   // you'll need to implement proper c'tor and d'tor fns
   CMoveNode() {}
   ~CMoveNode() {}

   BYTE m_XMove;
   BYTE m_YMove;
   CMoveNode *m_pNextSibling;
   CMoveNode *m_pFirstChild;
   CMoveNode *m_pParent;         // do you need this???

private:
   // private members
};

So, all in all a maximum of about (presuming 32-bit pointers), 3*32 + 2*8 = 96 + 16 = 112 bits per node = 14 bytes.

If you were to take the system to 4 levels deep, you'd have as a worst case scenario 9P4 nodes = 3024 nodes, * 14 = ~42K.

Hope that's useful, it's not something I've done before but I've read up on things like this. I don't proclaim to be an expert, but would also be interested to hear yours and others comments on these ideas. Maybe you could do an article when you've finished Big Grin | :-D

.

--
Andrew.

alex.barylski18-Feb-02 14:49

alex.barylski

18-Feb-02 14:49

I'm pretty sure recursion for a function with as much code as AI would be over kill.

char gameBoard[9] = {0,0,0,
                     0,0,0,
                     0,0,0};

BOOL IsWin() const
{
  char Piece = 'X';  //Look for X's first

  ...BEGIN loop
  //Is there a horizontal win
  if(gameBoard[0] == Piece && gameBoard[1] == Piece && gameBoard[2] == Piece) return TRUE;
  if(gameBoard[3] == Piece && gameBoard[4] == Piece && gameBoard[5] == Piece) return TRUE;
  if(gameBoard[6] == Piece && gameBoard[7] == Piece && gameBoard[8] == Piece) return TRUE;

  //Is there a vertical win
  ...
  //Is there a diagonal win
  ...

  Piece = 'Y';  //Loop once more to test for Y's
  ...END loop

  return FALSE;
}

This is something like what I did

Cheers

"An expert is someone who has made all the mistakes in his or her field" - Niels Bohr

John Cruz19-Feb-02 1:19

John Cruz

19-Feb-02 1:19

hey thank for the help... shoq, andrew, christian and hockeydude. Laugh | :laugh:

well, i think its not gonna be an over kill to use recursion because i am only implementing a Tictactoe board. so theres only 9 possible moves. but i think ill try do both recursion and the list idea. for recursion, ill make moves made and if the board is all occupied as my base case. as the general case it will just continue look at at possible moves - the moves done. for the list idea. ill try to immitate the deep blue idea by making a list of all the moves and putting it on a List or a stack. Wink | ;)

after i finish this ill try show you guys what i came up with... Big Grin | :-D