|
Hi,
My project need use a exe(statically linked MFC) load a dll(statically linked MFC),
and create a CDHtmlDialog in the dll.
I add a CDHtmlDialog to the DLL, "class DHtmlDialogTest : public CDHtmlDialog"
and create dialog at DLL's export function,
"DHtmlDialogTest.Create(IDD_DHtmlDialogTest , CWnd::GetDesktopWindow());"
But when excute DHtmlDialogTest.Create(IDD_DHtmlDialogTest , CWnd::GetDesktopWindow());
then it'll enter "dlgdhtml.cpp"'s function "CDHtmlDialog::OnInitDialog()"
at following code inside "CDHtmlDialog::OnInitDialog()"
--------------------------------------------------------------------------
if (m_pBrowserApp == NULL)
{
// create the control window
m_wndBrowser.CreateControl(CLSID_WebBrowser, NULL,
WS_VISIBLE | WS_CHILD, rectClient, this, AFX_IDC_BROWSER);
lpUnk = m_wndBrowser.GetControlUnknown();
if (FAILED(lpUnk->QueryInterface(IID_IWebBrowser2, (void**) &m_pBrowserApp)))
{
m_wndBrowser.DestroyWindow();
DestroyWindow();
return TRUE;
}
}
--------------------------------------------------------------------------
it will GPF which is get a unhandle exception when excute
if (FAILED(lpUnk->QueryInterface(IID_IWebBrowser2, (void**) &m_pBrowserApp)))
the m_wndBrowser is NULL and lpUnk is also NULL.
and,
1. I can create a CDialog in the same DLL above, only can't create a CDHtmlDialog.
2. If use a exe(dynamically link MFC) load a dll(Extension DLL use MFC),
it have no problem,
it can do the same thing, create a CDHtmlDialog in the dll perfectly.
Maybe because I can't load the resource?
Because I trce the call stack, when enter afxwin2.inl excute following code:
CDialog::Create(ATL_MAKEINTRESOURCE(nIDTemplate), pParentWnd);
ATL_MAKEINTRESOURCE(nIDTemplate) return a bad ptr,
Or is there any thing I need to be careful when use a exe(statically linked MFC) load a dll(statically linked MFC),
and create a CDHtmlDialog in the dll?
Thanks for your help~
-- modified at 3:02 Friday 30th June, 2006
|
|
|
|
|
Since the dialog is based on WebBrowser ActiveX control, I suppose you should enable ActiveX controls using this initialization function:
<pre><code>AfxEnableControlContainer();</code>
Put it in InitInstance handler of your CWinApp class in DLL.
Maybe some other initialization functions are required too.
I hope it helps.
|
|
|
|
|
Hi, Viorel,
After I tried put AfxEnableControlContainer();
it still GPF at same line.
Ryan
|
|
|
|
|
I know how to solve it,
use ::OleInitialize(NULL);
Thanks
|
|
|
|
|
I have the following problem:
I have a dll compiled with MinGW and a header file that describes the DLL exports. I want to use this DLL in a Visual C++ projects. Since i don't have a LIB file and i don't want to use LoadLibrary function i created a DEF file. The contents of the EXPORTS section of the DEF file i toke from DUMPBIN /EXPORTS [the_dll], so it should be ok.
After i created the lib file i included it in my Visual C++ project and the project has compiled successfully. But when i executed it, the application crashed. This application is working when compiled with GCC and the same DLL.
So is the problem somewhere in MinGW or what?
May be i must pass some special parameters to cl.exe or link.exe
Experts please help!
|
|
|
|
|
This has probably nothing to do with the fact that it's a dll. Start your debugger to see where the program crashes.
Cédric Moonen
Software developer
Charting control
|
|
|
|
|
Ok i will explain in more details:
So in the header file i have a definition of an abstract class wich has onlu pure virtual methods.
In the DLL i have only one export function which is returning a pointer to an instance of this class or more precisely some class that inherits this abstract class.
The program is something like:
<br />
SomeClass* c = ImportedMethod();<br />
c->someMethod(); <br />
It looke strange i know, the problem is that i don't hava the source of the DLL neither i have any more information about it. Only those header:
<br />
class SomeClass <br />
{<br />
public:<br />
virtual void someMethod() = 0;<br />
}<br />
extern "C"<br />
__declspec (dllexport) SomeClass * ImportedMethod();<br />
|
|
|
|
|
What is the value of c ? Does it point to a valid memory address ?
Otherwise your method could simply return NULL and you didn't check it.
Cédric Moonen
Software developer
Charting control
|
|
|
|
|
c isn't null. c contains a valid memory address and according to the Visual Studio debugger it contains the addresses of virtual functions.
The same source is working when compiled with GCC.
The difference may be is in the calling conventions or i don't know.
But there is no problem in the source.
The problem is in the dll that is used i think.
|
|
|
|
|
Hey guys
I was wondering why you use
typedef Handle ****Handle;
What does it exactly do?? 
I'm sorry guys, I didn't mean *** to be a pointer rather a dummyname
typedef Handle ObjectHandle
This is what I meant
-- modified at 11:10 Thursday 29th June, 2006
|
|
|
|
|
It's a pointer to a pointer to a pointer to a pointer to a Handle 
are you certain that's the way it is written ? because typedefing with the name name should not work.
In what context is it used ?
Maximilien Lincourt
Your Head A Splode - Strong Bad
|
|
|
|
|
It is used in RTI (Runtime Infrastructure)
|
|
|
|
|
Is this real code?
Quite a few indirections to itself!
Best,
Jun
|
|
|
|
|
Yeah it is actually used in RTI to call upon different federates
Im not sure if u are familiar with RTI.
What it does is, interacts with multiple simulations which the simulations interacting with each other. Its more like a third party.
|
|
|
|
|
A little bit, but I haven't done anything seriously yet. I don't like SOM/FOM stuff.
Are you developing a HLA federation now? It's quite a lot of effort from my understanding. Which vender's RTI are you using, VTC, MAK, PITCH, etc?
Best,
Jun
|
|
|
|
|
Some of us use it in very rare situations like this:
struct Handle
{
};
struct Handle2
{
typedef Handle ****Handle;
};
-- modified at 10:41 Thursday 29th June, 2006
It means that within Handle2, Handle will be a pointer to a pointer to a pointer to a pointer which points to a value of first structure.
|
|
|
|
|
|
Probably Handle name is used for some convenience, or maybe to hide the original meaning of Handle.
|
|
|
|
|
Yeah, that's correct. However, I'm wondering:
- where need so many levels of indirection?
- wouldn't be better to change anyhow the defined type name to be clearer for anybody what's there?
Ovidiu
|
|
|
|
|
I think we should ask the authors of RTI, where such weird constructions are used, according to Jay03 message.
|
|
|
|
|
It is a very poor way to declare a new type. The reason it is poor is because of the naming. The goal it is trying to accomplish is to declare a new type that is a pointer-to-a-pointer-to-a-pointer-to-a-pointer-to-a-Handle. A better way to write it would be either of the following:
Option 1:
typedef RedirectedHandle ****Handle;
Option 2:
namespace MyStuff
{
typedef Handle **** ::Handle;
}
An even better solution would be to not use that many levels of indirection (who needs 4 levels of indirection for a Handle?) and to use smart pointers.
If you decide to become a software engineer, you are signing up to have a 1/2" piece of silicon tell you exactly how stupid you really are for 8 hours a day, 5 days a week
Zac
|
|
|
|
|
A little bit better for everybody... :thumb:
Ovidiu
|
|
|
|
|
void CMyButton::PreSubclassWindow()
{
UINT nBS;
nBS = GetButtonStyle(); // GetStyle & 0xFF == all bs_* OR operator ?
m_nTypeStyle = nBS & BS_TYPEMASK; // GetStyle & 0xFF & 0xF ==? == all bs_* OR operator ?
if (nBS & BS_CHECKBOX)
m_bIsCheckBox = TRUE;
if (m_nTypeStyle == BS_DEFPUSHBUTTON) // why not use m_nTypeStyle & BS_DEFPUSHBUTTON
{
m_bIsDefault = TRUE;
}
ASSERT(m_nTypeStyle != BS_OWNERDRAW);// why not use !(m_nTypeStyle & BS_OWNERDRAW)
ModifyStyle(BS_TYPEMASK, BS_OWNERDRAW, SWP_FRAMECHANGED);
CButton::PreSubclassWindow();
}
LRESULT CMyButton::OnSetStyle(WPARAM wParam, LPARAM lParam)
{
UINT nNewType = (UINT)(wParam & BS_TYPEMASK); // why not use GetButtonStyle() & BS_TYPEMASK -> wParam & 0xFF & 0xF?
if (nNewType == BS_PUSHBUTTON) // why not use nNewType & BS_PUSHBUTTON? (if SetButtonStyle(BS_OWNERDRAW | BS_PUSHBUTTON)
{
m_bIsDefault = FALSE;
}
return DefWindowProc(BM_SETSTYLE,
(wParam & ~BS_TYPEMASK) | BS_OWNERDRAW, lParam); // why in here remove BS_TYPEMASK style?
}
thanks.
|
|
|
|
|
|
Let me guess!
It's the last Code Inspection Report for your project. Right? 
Ovidiu Cucu
Microsoft MVP - Visual C++
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
