|
I read this article about calling conventions:
http://www.codeproject.com/cpp/calling_conventios_demystified.asp
Something is not clear for me:
1-) I see that function overloading is only possible with thiscall. Right? Is this true?
"Thiscall is the default calling convention for calling member functions of C++ classes"
I can set Visual C++ Compiler Options with /Gd, /Gr, /Gz (Calling Convention) . __cdecl, __fastcall, or __stdcall .
2-) But this doesnt affect C++ member class functions calling convention type, i mean thiscall. They are always thiscall. Right?
3-) In win32 programming i see all of functions declare with stdcall. So beacuse of that win32api is not object oriented. I mean can not use function overloading. But why didnt they design win32api with thiscall?
Can you please verify my conclusions. Am i wrong or right?
Thank you.
|
|
|
|
|
sawerr wrote: Am i wrong or right?
Yes. C (stdcall) versus C++ (thiscall), they are not the same.
led mike
|
|
|
|
|
sawerr wrote: But why didnt they design win32api with thiscall?
That would have been sweet for us C++ programmers, and .NET is a BIG step that direction, but I
imagine since the OS APIs need to be callable from any language (including non-OOP ones) they had
to use a common calling convention.
"Great job, team. Head back to base for debriefing and cocktails."
(Spottswoode "Team America")
|
|
|
|
|
"thiscall" is not a keyword, it means "the function has this as its first parameter even though it's not written that way". A member function can be __stdcall - when you write an implementation of a COM interface in C++, the methods are __stdcall (it's hidden in the STDMETHODIMP macro).
Function overloading is done by the compiler giving each overload different internal names. This is unrelated to calling conventions.
Most Win32 APIs use __stdcall (or as it was originally called, PASCAL). Only a few like wsprintf() use __cdecl. APIs can't be overloaded because that would break compat with Pascal and C and other languages that don't know about overloading or mangled names.
|
|
|
|
|
Thanks for all answers.
Micheal Dunn you told me:
"Function overloading is done by the compiler giving each overload different internal names. This is unrelated to calling conventions."
Yes i tried a piece of code
#include <iostream><br />
using namespace std;<br />
<br />
class CRectangle {<br />
int x, y;<br />
public:<br />
void set_values (int,int);<br />
int _stdcall area () {return (x*y);}<br />
int _stdcall area (int x ) {cout << x; return 0;}<br />
};<br />
<br />
void CRectangle::set_values (int a, int b) {<br />
x = a;<br />
y = b;<br />
}<br />
<br />
int main () {<br />
CRectangle rect;<br />
rect.set_values (3,4);<br />
cout << "area: " << rect.area();<br />
rect.area(8);<br />
return 0;<br />
}
It works. I tried both _stdcall and _cdecl there is no problem.
But i dont understand here. Why do they write:
"For C calling convention (__cdecl): Function name is decorated by prefixing it with an underscore character '_' .
For Standard calling convention (__stdcall): Function name is decorated by prepending an underscore character and appending a '@' character and the number of bytes of stack space required."
For cdecl Compiler generate function name for area():
?area@CRectangle@@QAAHXZ
?area@CRectangle@@QAAHH@Z
So function name decoration information is not true for member functions. Because function names both 2 area func must be "_area". Right?
If function overloading is done by compiler why is there a specification for calling conventions? Why is there an information about function name decoration? VC++ doesnt obey this rule.
Can you please explain here please.
Thanks for answers
|
|
|
|
|
The calling convention determines how the parameters are put on and cleaned off the stack. Overloading is done strictly by generating unique mangled names (that's what the funky-looking ?area@CRectangle... names are). The name decorating rules you quoted hold true for exported and non-overloaded functions. As soon as you introduce overloading, you get the long mangled names.
|
|
|
|
|
I need to implement a window that is suitable for things like a combo box's dropdown window, or a menu.
- I want it to popup (be top most) with out recv'ing focus.
- I want it to ignore mouse activates
- I want it to be closed when a WM_LBUTTONUP is recv'd by the window, OR a WM_LBUTTONDOWN happens on ANY other window.
I have items 1 and 2 working fine. However I'm not sure of the best way to implement the third one. Is installing a windows hook the best way? If so, which hook type do I want? There is a specific mouse hook type, but it doesn't look your call back gets enough info - i.e. you can't tell if it's a mouse down, etc.
|
|
|
|
|
Dear All;
I want to capture a screen shot of a web browser control (m_webBrowser) which is hosted in a dialog box.
Here is the code which i used to do that but its not working properly (ie. capturing other windows) !
<br />
RECT rc;<br />
LPOLEWINDOW pOWin = 0;<br />
HWND hWnd;<br />
m_webBrowser.GetControlUnknown()->QueryInterface(IID_IOleWindow, (LPVOID*)&pOWin);<br />
pOWin->GetWindow(&hWnd);<br />
HWND hwFirst = ::GetWindow( hWnd, GW_CHILD );<br />
HWND hwSecond = ::GetWindow( hwFirst, GW_CHILD );<br />
::GetWindowRect (hwSecond,&rc); <br />
HDC hDC = ::GetDC(0);<br />
HDC memDC = ::CreateCompatibleDC ( hDC );<br />
HBITMAP memBM = ::CreateCompatibleBitmap ( hDC, rc.right-rc.left, rc.bottom-rc.top );<br />
::SelectObject ( memDC, memBM );<br />
::BitBlt( memDC, rc.left, rc.top, rc.right-rc.left, rc.bottom-rc.top , hDC, 0, 0 , SRCCOPY );<br />
<br />
int size = 3 * ( (rc.right-rc.left) * (rc.bottom-rc.top) );<br />
BYTE *lpBits = new BYTE[size];<br />
<br />
::GetBitmapBits( memBM, size, lpBits );<br />
<br />
char* imageName;<br />
imageName =(char*)"2.jpg";<br />
SaveBitmap(imageName, memBM);
<br />
delete [] lpBits;<br />
::DeleteObject(memBM);<br />
::DeleteObject(memDC);<br />
::ReleaseDC( 0, hDC )<br />
Can someone please help me? Thank you
llp00na
|
|
|
|
|
Maybe you need to use EnumChildWindows() instead of GetWindow() to find the appropriate window.
Just a guess since I have no idea what your window hierarchy looks like 
Mark
"Great job, team. Head back to base for debriefing and cocktails."
(Spottswoode "Team America")
|
|
|
|
|
thanx for your reply;
I only have a dialog box (main window) which hosts the webBrowser control.
llp00na
|
|
|
|
|
So where is this code being called from?
"Approved Workmen Are Not Ashamed" - 2 Timothy 2:15
"Judge not by the eye but by the heart." - Native American Proverb
|
|
|
|
|
The code is being called from a button click handler. (The user clicks a button "Grab Screen").
llp00na
|
|
|
|
|
llp00na wrote: The code is being called from a button click handler.
Then just use the web control's GetWindowRect() method (instead of calling GetWindow() twice).
"Approved Workmen Are Not Ashamed" - 2 Timothy 2:15
"Judge not by the eye but by the heart." - Native American Proverb
|
|
|
|
|
I have tried what you suggested but it stil gives me the same result (ie. similar to the one i got when i called getWindow() twice)
It grabs part of the MVStudio, part of the dialog which hosts the web browser and the remaining of the screen shot is painted in black.
Any other ideas would be appreciated
llp00na
|
|
|
|
|
Continuing from Mr Crow's question...
...and which window's pixels is it grabbing?
Mark
"Great job, team. Head back to base for debriefing and cocktails."
(Spottswoode "Team America")
|
|
|
|
|
its grabbing part of the Microsoft VC++, parts of the dialog window (and not the web browser at all). The remaining of the screenshot is all black.
llp00na
|
|
|
|
|
The black is probably because the bitmap bits size (in bytes) is calculated wrong.
Try something like
BITMAP bitmap;
::GetObject(memBM, sizeof(BITMAP), &bitmap);
long size = bitmap.bmWidthBytes * bitmap.bmHeight;
BYTE *lpBits = new BYTE[size];
...
or
BITMAP bitmap;
::GetObject(memBM, sizeof(BITMAP), &bitmap);
long size = (((bitmap.bmWidth * (long)bmBitsPixel + 15L) & (~15L)) / 8L) * bitmap.bmHeight;
BYTE *lpBits = new BYTE[size];
...
"Great job, team. Head back to base for debriefing and cocktails."
(Spottswoode "Team America")
|
|
|
|
|
I have tried your suggestions and i still have the same problem. No difference what so ever !
llp00na
|
|
|
|
|
Sorry I should have looked more closely at your code.
Try changing your BitBlt() call from
::BitBlt( memDC, rc.left, rc.top, rc.right-rc.left, rc.bottom-rc.top , hDC, 0, 0 , SRCCOPY );
to
::BitBlt( memDC, 0, 0, rc.right-rc.left, rc.bottom-rc.top , hDC, rc.left, rc.top , SRCCOPY );
"Great job, team. Head back to base for debriefing and cocktails."
(Spottswoode "Team America")
|
|
|
|
|
Oh thanx so much man, you are a genius;).
your code captures a screen shot of the visible area of the web browser. Do you know how to capture all area of the web browser -including the invisible parts (That is when the user has to scroll down to view the rest of the page). Because this is when i need to do.
llp00na
|
|
|
|
|
Since your grabbing pixels from the display, you won't be able to get the non-visible pixels
that way.
Maybe there's an overridable method in the browser control where you can force the HTML to be
rendered to a dc of your choice instead of the screen...?
Mark
"Great job, team. Head back to base for debriefing and cocktails."
(Spottswoode "Team America")
|
|
|
|
|
I was thinking if i can get a handle of the web browser then it should be possible to capture all area of the web browser. Do you know how to retrieve the handle of the web browser?
llp00na
|
|
|
|
|
Hello,
I am a newbie with VC++ and MFC and I try to understand how CTreeCtrl works.
Could you please tell me how to use lparam in TVITEM.
I've found nothing in the MSDN and I also searched in this forum. I've read that someone used it to store a LPSTR.
Does it mean that I can use it to store my own types such pointers, structures etc..?
many thanks for your help.
|
|
|
|
|
Arris7 wrote: Does it mean that I can use it to store my own types such pointers, structures etc..?
Yes, you can do it.
In our case, in the tree, each leaf points to its corresponding object via the pointer that is stored in the lparam member of the TVITEM :
In this example, when I insert a leaf in the tree, I will assing the pointer to lparam, you must not forget to add the TVIF_PARAM to the mask. ( note that the TVITEM in this case is a member of the TVINSERTSTRUCT struct )
TVINSERTSTRUCT tv;
ZeroMemory (&tv, sizeof ( tv ) );
tv.hParent = hParent;
tv.hInsertAfter = TVI_LAST;
tv.item.mask = 0;
tv.item.mask = TVIF_IMAGE | TVIF_SELECTEDIMAGE | TVIF_TEXT | TVIF_PARAM;
tv.item.iImage = IMAGE_INDEX;
tv.item.iSelectedImage = IMAGE_INDEX;
tv.item.pszText = (LPTSTR)(LPCTSTR)sText;
tv.item.lParam = (DWORD)pPointerToObject;
HTREEITEM h = InsertItem( &tv );
|
|
|
|
|
Thnks for your explanation Maximilien, it sounds clear now.
please copuld you answer a last question before I go on into coding.
I use the following structure that will contain additional information related to my leaves
struct NodeInfo
{
CString Path;
CString GroupName;
CString Type;
DWORD Category;
};
so if I want to attach this structure to my leaf, can I write
tv.item.lParam = NodeInfo* moreinfo;
or is it better to write
NodeInfo* moreinfo = new NodeInfo; // Do not forget to delete
tv.item.lParam = moreinfo;
Thnks again for your help
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
