|
Alan Balkany wrote: This is a known quirk in C++: For static variables in a class, you need to repeat the declaration in the .cpp file.
It's not a quirk, and you're not re-declaring the variable.
The one-and-only static member variable instance needs to exist
somewhere, so you have to define that instance. 
Mark
Mark Salsbery
Microsoft MVP - Visual C++
![Java | [Coffee]](https://codeproject.global.ssl.fastly.net/script/Forums/Images/coffee.gif)
|
|
|
|
|
But this is counterintuitive; you don't have to redeclare the non-static class members.
I got burned by this quirk years ago. I've seen several other people run into this same problem on this forum over the years. There's no good reason for this behavior.
|
|
|
|
|
Alan Balkany wrote: you don't have to redeclare the non-static class members.
A declaration is not an instance. You are NOT declaring static member variables twice.
You declare the variable once, and you also must provide an instance of the variable.
Non-static members are not the same - every existing object (instance) of the type gets its
own instance of the variable.
Alan Balkany wrote: There's no good reason for this behavior.
You'll have to ask Bjarne for the good reason
Mark
Mark Salsbery
Microsoft MVP - Visual C++
|
|
|
|
|
Yet people continue to get burned by this "feature".
I see bad design every day.
|
|
|
|
|
"burned"? How so? It's just a simple linker error - a LOT easier
than tracking down a runtime error
Mark Salsbery
Microsoft MVP - Visual C++
|
|
|
|
|
Burned: You write code that the compiler accepts, but it doesn't work. You waste hours or days until you (somehow) learn the gimmick that gets by the quirk.
Good language design minimizes instances of this. Stroustrup isn't infallible, and the fact that he implemented this doesn't automatically make it good.
|
|
|
|
|
Right. I never said it was good. Every language has its trade-offs.
This particular case has always been the case. Had the C++ designer(s) let the
compiler choose where to create the instance, there's would be a whole lot
of us whining because we don't have control over the storage of our variables.
I use C++ for the low level control. I understand how objects exist and where, and
I take advantage of the control I have over that. So to me it is not counter-intuitive
to have to explicitly state where an object is stored. For times I don't care, I'll use
a higher level language - C# works just fine
The whole "static" thing in itself is kind of an abomination itself - mixing linkage with compilation.
Cheers,
Mark
Mark Salsbery
Microsoft MVP - Visual C++
|
|
|
|
|
Since there's only one instance of a static variable, it doesn't matter where it's stored.
It's counterintuitive because neither instance variables NOR "ordinary" static variables require two declarations, and the thousands of people who have had their time wasted by this quirk support this assertion.
|
|
|
|
|
Alan Balkany wrote: and the thousands of people who have had their time wasted by this
Thousands? I thought EVERYONE went through it at least once
I get where you're coming from - I just do not agree that there's two declarations. There's
a distinct difference between declaration and an actual instance of an object.
Mark Salsbery
Microsoft MVP - Visual C++
|
|
|
|
|
This linked fine for me:
class Test : public CDialog
{
DECLARE_DYNAMIC(Test)
public:
static int cnt;
};
class Test1
{
void GetData()
{
int cnt = Test::cnt;
}
};Can you compare it to what you have?
"Love people and use things, not love things and use people." - Unknown
"To have a respect for ourselves guides our morals; to have deference for others governs our manners." - Laurence Sterne
|
|
|
|
|
Hi David,
I fix the problem. Please refer the above.
|
|
|
|
|
That shouldn't link, at least not if you actually use one of those classes
Mark Salsbery
Microsoft MVP - Visual C++
|
|
|
|
|
Mark Salsbery wrote: ...at least not if you actually use one of those classes
I didn't.
"Love people and use things, not love things and use people." - Unknown
"To have a respect for ourselves guides our morals; to have deference for others governs our manners." - Laurence Sterne
|
|
|
|
|
Anyone know of an x64 PKWare compatible compression library (or a library with source code that would allow me to build the x64 version). I currently use an old version of the PKWare Data Compression Library that allows blocks of data to be compressed and restored but there isn't a 64bit version and PKWare don't list 64 bit versions of anything on their website.
|
|
|
|
|
Hi,
I have created a LIB using Win32 Static Library in which i have overriden the CWnd ::Create method.
when i use this LIB in MFC application to create the Window its giving ASSERSETION
due to parent window handle is becoming NULL.
|
|
|
|
|
Window handles don't just "become NULL".
The problem could be anything from mismatched or multiple MFC libraries
to errors in your code. Have you tried to debug it?
Mark
Mark Salsbery
Microsoft MVP - Visual C++
|
|
|
|
|
Hi,
I have created the Win32 Static Library which has MFC Support and Using Precompiled Header
(This LIB has been created by selecting Win32 Static Library from MFC Project Option )
I have added one class to library which is derived from CWnd, In Derived Class i have overriden the
Create()method to create the window.
At Application side
I have MFC SDI application which has multiple splitters where each pane of splitter corrsoponds to one class.
One of the class wants to create the window using the LIB that i have created.
when i give call to the Create from application side it is giving assertion on
ASSERT(afxCurrentInstanceHandle != NULL)
To Test LIB,i have tried for following things
I have created the TestApp using MFC SDI and while generating the Application CTestView is derived from the CFormView.
here call to Create of LIB is working fine.
|
|
|
|
|
Hello,
I need to parse a data file which contains reals.
The decimal separator use is always the dot '.'
I try to extract the values using a stringstream to get the result into a double.
As you probably know, stringstream use the current locale to determine the decimal seperator to use...
and that is my problem !
How can I tell to my stringstream to use the dot separator regardless the current locale ?
I can not find a "simple" snippet to do this.
Any idea ?
|
|
|
|
|
jeremie.delaitre wrote: Any idea ?
I don't know the specific answer. Based on my limited knowledge of STL I would think you could create a specific locale and make the stream use it. I think streams have like imbue() function or something to make them use specific locales.
led mike
|
|
|
|
|
Yes there is an imbue method and I can, in theroy, create my own.
Indeed, the documentation I have on it is vague so I wonder if a simpler solution exists.
|
|
|
|
|
jeremie.delaitre wrote: so I wonder if a simpler solution exists.
Depends on your definition of simpler. You could certainly parse the data yourself, perhaps using Regular Expressions.
led mike
|
|
|
|
|
Dear All,
My requirement is to change the bitmap of the button on mouse move and mouse over. I have used TrackMouse event to do this. But I have problem in this, that is, when I place a mouse over the button the bitmap is changed and again when I take mouse over the button the bitmap is again changed. This looks bad due to painting and re-painting continously. Please advice me a good way. The code is below :
void Test::OnMouseMove(UINT nFlags, CPoint point)
{
TRACKMOUSEEVENT trackmouseevent;
if(!click2)
{
trackmouseevent.cbSize = sizeof(trackmouseevent);
trackmouseevent.dwFlags = TME_LEAVE;
trackmouseevent.hwndTrack = GetSafeHwnd();
trackmouseevent.dwHoverTime = HOVER_DEFAULT;
_TrackMouseEvent(&trackmouseevent);
}
GetDlgItem (IDC_BUTTON3)->GetWindowRect (&rect);
ScreenToClient (&rect);
if ((!rect.PtInRect(point)))
{
bool val=but2.LoadBitmaps(IDB_BITMAP23,0,IDB_BITMAP23,0);
but2.Invalidate();
but2.UpdateWindow();
}
GetDlgItem (IDC_BUTTON2)->GetWindowRect (&rect);
ScreenToClient (&rect);
if ((!rect.PtInRect(point)))
{
bool val=input.LoadBitmaps(IDB_BITMAP22,0,IDB_BITMAP22,0);
input.Invalidate();
input.UpdateWindow();
}
CDialog::OnMouseMove(nFlags, point);
}
LONG Test::OnMouseLeave(WPARAM , LPARAM)
{
POINT point;
GetCursorPos(&point);
ScreenToClient(&point);
CRect rect;
GetDlgItem (IDC_BUTTON2)->GetWindowRect (&rect);
ScreenToClient (&rect);
if (rect.PtInRect(point))
{
bool val=input.LoadBitmaps(IDB_BITMAP57,0,IDB_BITMAP57,0);
input.Invalidate();
input.UpdateWindow();
click2=false;
}
GetDlgItem (IDC_BUTTON3)->GetWindowRect (&rect);
ScreenToClient (&rect);
if (rect.PtInRect(point))
{
bool val=but2.LoadBitmaps(IDB_BITMAP58,0,IDB_BITMAP58,0);
but2.Invalidate();
but2.UpdateWindow();
click2=false;
}
|
|
|
|
|
Unless I'm mistaken, you're reloading the button's bitmap each and every time the mouse sends a move event. Why don't you only draw the bitmap when it changes?
You could either keep a track of the mouse position last move, and compare the last pos to the current pos, to see if the mouseOver status has changed for either button.
Or, you simply could add some flags to hold the current mouseOver status of each button. Comparing the current status with the last to determine if the bitmap needs to be changed to reflect the mouse position.
|
|
|
|
|
Hi
I am creating plugin for VS2005 in which i want to handle the Click event of Window->New Vertical Tab. i.e. Whenever user click on plugin icon, "New Vertical Tab's events should call. So Is there any API for handling above events.
Regards
aaa
|
|
|
|
|
HI,
in CString Find Function can i use variable insted of fixed text.
Just for example .
CString str = _T("ABCD edbd CDE ~");
int nPosition = str.Find("~",0);
instead of passing " ~ " can i use any varible ?
Just like this CString EndDelimeter = _T("~");
int nPosition = str.Find(EndDelimeter ,0);
Because user can change any special charecter as End delimeter.
Plz advice me
Preethi
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
