|
I have a problem which I am finding myself unable to figure out.
I have a fairly simple CSocket-derived class, which does not much beyond overriding OnReceive so that it saves all the received buffer, and then calling a function of a Pointer to a thread class of mine, which does what it will with the buffer.
At the moment I have a server and a client, both of which have a thread for receiving and a thread for sending through these sockets. The general communication type is thus - client sends a request, server processes and sends some sort of a reply. The messages to be sent in-process are delivered through inter-thread messaging.
The problems occurs mainly while debugging, but it can happen while just running the applications - at some point, if a number of request are "fired" rapidly, the connection get messed up in a rather odd way:
Clients sends successfully, then Server receives, does it's processing, goes on to send a reply, the send operation goes without errors, and yet the other side never receives the message. I've traced all the receive calls on my socket class, and not once there was an error.
So basically, everything indicates that the connection is alive, and yet it never goes through in one direction.
The messages are of varied sizes, but when in debug, 2 140-bytes requests and 160-bytes replies kill the communication.
|
|
|
|
|
Does the error occur for any of the unit tests?
What sort of coverage do your unit tests supply?
|
|
|
|
|
Are you receiving just the number of bytes you're expecting or receiving all the bytes sent?
If the latter, are you parsing the bytes into individual "messages" or assuming it's one message and losing the rest?
Mark
Mark Salsbery
Microsoft MVP - Visual C++
![Java | [Coffee]](https://codeproject.global.ssl.fastly.net/script/Forums/Images/coffee.gif)
|
|
|
|
|
I am receiving the number bytes I'm expecting, first by receiving a buffer size parameter,and then the rest of the message.
|
|
|
|
|
You mentioned another thread.
OnReceive() occurs on the thread that creates the socket (through a hidden HWND).
Is this your UI thread?
Seems like something's causing you to lose FD_READ notifications from the socket.
I can only guess without seeing the code.
Mark
Mark Salsbery
Microsoft MVP - Visual C++
|
|
|
|
|
CSocket is also notorious for being a bug-riddled piece of code. You may want to switch to CAsyncSocket or roll your own.
|
|
|
|
|
How do you add text to the title bar of an application using Microsoft Visual Studio 2003 .NET Visual C++?
|
|
|
|
|
SetWindowText() or WM_SETTEXT.
Mark
Mark Salsbery
Microsoft MVP - Visual C++
![Java | [Coffee]](https://codeproject.freetls.fastly.net/script/Forums/Images/coffee.gif)
|
|
|
|
|
|
This application has a mainframe window and like an older version of Microsoft Word documents, it can have multiple files open (MDI). If you tile the windows, the active window's file name title shows in the mainframe's. The format the header or title bar has is as follows:
[Application Name] - [Filename.extension]
I want to modify it like this:
[Application Name] - [Filename.extension] - [My Text Added Here]
The problem is I do not have a handle to this section such as SetTitle() or SetName() and I am thinking I need to add a section to have such a handle. However, this is dealing with MFC.
|
|
|
|
|
GetWindowTextLength()/GetWindowText() or W<_GETTEXT/WM_GETTEXTLENGTH can be used to get
the current title.
Append your new text and set the new text.
For MFC you can use something like
CString titlestr;
MainWnd->GetWindowText(titlestr);
titlestr += _T(" - [My Text Added Here]");
MainWnd->SetWindowText(titlestr);
Mark
Mark Salsbery
Microsoft MVP - Visual C++
|
|
|
|
|
Again[^]
----------------------------
286? WOWW!
|
|
|
|
|
Are you simply wanting a 2D array?
"A good athlete is the result of a good and worthy opponent." - David Crow
"To have a respect for ourselves guides our morals; to have deference for others governs our manners." - Laurence Sterne
|
|
|
|
|
Thanks atlast got a reply.
No I'm not simply wanting a 2D Array.
An array is a group of memory locations that we reserve, Ok in the below statement, I've reserved. 6 Memory locations that has the capability to store 6 integers. And I've got a pointer pointing to that.
int* pInt = new int[2*3];
The dimension. It's how we look at the reserved memory location.
The memory as of now, looks like :
[] [] [] [] [] []
^
Incrementing the pInt's value would make it go like :
[] [] [] [] [] []
-+^
[] [] [] [] [] []
---++^
This is linear. We can do the same with pInt[0],[1]..etc. Ok this about looking at the memory locations in 1D.
Now I want to look at it in 2D.
The same memory location.
I would want to initialize a 2D pointer to the same memory location.
Where I should be able to mention 2 indices. You may take it as 2x3 or 3x2
when you say 2x3,
That says, 2 rows and 3 elements in each row.
[] [] []
[] [] []
Now tell me how should declare a pointer to the memory location that it would access them as 2D.
int* pInt = new int[2*3];
int(** ptr2D)[2][3] = new int*[2]; <--something like this. NOT SURE.
If we are gonna look at the same memory locations as 3x2,
then it becomes.
[] []
[] []
[] []
Here ptr2D[2][0].. would make sense. Because, The number of rows, 0th row, 1th row, 2nd row. Has 3 rows. But in the former case, it would crash because we were looking at it as 2 rows with 3 columns.
My question makes sense?
----------------------------
286? WOWW!
|
|
|
|
|
_8086 wrote: Now tell me how should declare a pointer to the memory location that it would access them as 2D.
int **p = new int*[FirstDimension];
for(int i = 0; i < FirstDimension; i++)
{
p[i] = new int[SecondDimension];
for(int j = 0; j < SecondDimension; j++)
{
p[i][j] = 0;
}
}
"A good athlete is the result of a good and worthy opponent." - David Crow
"To have a respect for ourselves guides our morals; to have deference for others governs our manners." - Laurence Sterne
|
|
|
|
|
Let me see if this satisfies my requirement. Anyway thanks for your reply.
----------------------------
286? WOWW!
|
|
|
|
|
It does not matter. pInt[2][3] is the same as pInt[3][2].
|
|
|
|
|
Yes, but the question is different.
----------------------------
286? WOWW!
|
|
|
|
|
Rage wrote: It does not matter. pInt[2][3] is the same as pInt[3][2].
Oh, you've said like this?
Ok, take up the first one., pInt[2][3];
Try Printing the element pInt[3][0]."Boom" !??!. How can this be eqivalent to pIntp[3][2].
The dimension is different and that's what the question is all about. Looking the same block of memory with different dimensions.
----------------------------
286? WOWW!
|
|
|
|
|
No, you are reserving 6 memory locations. How you read them does not matter.
|
|
|
|
|
I don't think you can do what you're trying to do.
If you can, I'd like to see the syntax.
For a 2 dimension array like this:
int myarray[2][3];
myarray is not a pointer. You can't cast an int pointer to the type of myarray (AFAIK).
Mark
Mark Salsbery
Microsoft MVP - Visual C++
|
|
|
|
|
I have a dialog app with a function similar to this:
void CMyDlg::OnButton1()
{
//This bit works
WinExec("cmd.exe /c xcopy /s "+myfolder+" "+target_folder+" >>mylog.txt&exit",SW_HIDE);
// This is what i want to do:
// if (all files have been copied and cmd.exe exits then display the msgbox, if not, wait until filecopy is done and then display the msgbox)
// {
AfxMessageBox("Done! All files copied to target folder.",MB_OK|MB_ICONINFORMATION);
// }
}
The way the function works right now, the msgbox displays too early, and not after all the files are copied.
I guess i need some kind of return value, and then if its true/false then display msgbox.
How to solve this problem, any suggestions? Is this "doable" in any easy way?
Createprocess? If so, how would the Createprocess code look like?
thx!
|
|
|
|
|
Google for "CProcessWait". It used to be in an article here on CP, but PJ Naughter or Arendt (sorry do not know anymore) put it in a utility library.
|
|
|
|
|
rolfhorror wrote: Createprocess? If so, how would the Createprocess code look like?
Call CreateProcessEx() followed by WaitForSingleObject().
"A good athlete is the result of a good and worthy opponent." - David Crow
"To have a respect for ourselves guides our morals; to have deference for others governs our manners." - Laurence Sterne
|
|
|
|
|
In my program, I deliberately did not specify a background color in my WNDCLASSEX struct. As a result, I must paint it manually. My WNDPROC simply passes off all its arguments to an identical (static) function in a Controller class I created. The Controller class looks up the relevant HWND in a list and routes to the appropriate message handler for each object via a pointer GuiObject pointer.
My problem is the WM_ERASEBKGND message. The device context is contained in the WPARAM parameter, but when I try to pass WPARAM to my object's message handler, I get a linker error:
"LNK2001: unresolved external symbol "protected: virtual int __thiscall GuiObject::OnEraseBackground(void)" (?OnEraseBackground@GuiObject@@MAEHXZ)"
As far as I can tell, this only happens with Windows data types. I can pass ints and other things in with no problem.
Here is the code:
LRESULT Controller::MessageDispatcher(HWND hWnd, UINT msg, WPARAM wParam, LPARAM lParam)<br />
{<br />
for(current=head; !current->hWnd; current=current->next)<br />
if(current->next=NULL) break;<br />
<br />
switch(msg)<br />
{<br />
case WM_ERASEBKGND:<br />
current->OnEraseBackground((HDC) wParam);<br />
break;<br />
...<br />
}<br />
Can anyone tell me what I'm doing wrong?
Thanks in advance
-- modified at 11:32 Monday 6th August, 2007
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
