Thread: Problem trying to retrieve data from combo box! (strange issue)

Just a small update. I removed the setRefresh call from the initiation and left only the call to it inside the if ((LOWORD(wParam) == IDC_COMBO1) && HIWORD(wParam) == CBN_SELCHANGE) active. If I run the program once it produces garbage value. If I run it again in the same window it all works. I even tried delegating the dialog box procedure call to another function and that still doesn't help.

In other words, I have to run the dialog box twice in the same window to get it to work. I'm wondering if I've done something wrong in the WndProc window message handling function that calls the dialog box.

I'll add any more info as I find it. Thanks.

I don't see how this could be causing your issue, but the only thing I can see that you are definitely doing wrong is how you calculate the refresh rate. A refresh rate can be fractional, hence Numerator and Denominator, so when dividing those values you need to convert them to floating point first, and store the result as floating point, something like:

Code:

double Hz = double(Numerator) / Denominator; // if either operand is floating point then the other is also converted to fp

So loading the data would go something like:

Code:

for (int i = 0; i < count; i++)
{
    UINT n = modeList[i].RefreshRate.Numerator;
    UINT d = modeList[i].RefreshRate.Denominator;
    if (!mSetRefresh || double(n) / d == 60.0)
    {
        mResolutionPair.push_back(modeList[i].Width);
        mResolutionPair.push_back(modeList[i].Height);
        mResolutionPair.push_back(n);
        mResolutionPair.push_back(d);
    }
}

I've removed some code duplication also.
Some more duplication removal:

Code:

        if (LOWORD(wParam) == IDC_CHECK1)
        {
            D3DObj.mSetRefresh = (checkBox == BST_CHECKED);
            D3DObj.LogResolution(D3DObj.mdxgiFactory);
            SendMessageW(GetDlgItem(hDlg, IDC_COMBO1), CB_RESETCONTENT, 0, 0);
            setResolution(hDlg);
            return (INT_PTR)TRUE;
        }

Another thing is that your identifier _SKIP is technically not allowed. Identifiers starting with an underscore are reserved for the implementation in the global scope (but they can be used in local scopes). Identifiers starting with an underscore followed by an uppercase letter, or identifiers containing a double underscore anywhere(!), are reserved for the implementation in all scopes. So you should either just say SKIP or perhaps use something like SHOW_DEBUG_INFO. Also, you probably shouldn't make a second call to SendMessage for the debug part since that means you've sent two messages for debugging while only sending one for the regular code, which could make a difference in the result (considering your problem).

Code:

//#define SHOW_DEBUG_INFO // uncomment to show info
 
void setRefresh(HWND hDlg)
{
    WPARAM nI = ComboBox_GetCurSel(GetDlgItem(hDlg, IDC_COMBO1));
    D3DClass::resData* _resData = reinterpret_cast<D3DClass::resData*>(
        SendMessageW(GetDlgItem(hDlg, IDC_COMBO1), CB_GETITEMDATA, (WPARAM)nI, 0));
 
#ifdef SHOW_DEBUG_INFO
    std::wstring _string = std::to_wstring(_resData->w) + L" "
        + std::to_wstring(_resData->h)
        + L"\n" + std::to_wstring(_resData->Hz) + L"\n\n";
    OutputDebugString(_string.c_str());
#endif
 
    D3DObj.mRefreshRate = _resData->Hz;
    SetWindowTextW(GetDlgItem(hDlg, IDC_EDIT3), std::to_wstring(_resData->Hz).c_str());
}

BTW, on x86_64, LPARAM is a signed 64-bit integer while WPARAM is an unsigned 64-bit integer. "long" on windows is only 32-bit, while "long long" is 64-bit, both of them signed, of course, unless you also say "unsigned".

Once you've done that, upload the entire thing as a zip file somewhere and post a link here. Also describe exactly what the problem seems to be with the current version.

Hi there,

Here's the link to onedrive where I've uploaded it. Rather a big file i don't know if that's because of all the directX stuff in the includes. I may have included stuff I didn't need to but I'm not sure how to avoid that.

https://1drv.ms/u/s!AmIumtq4Ent9hjWP...VRgM1?e=zufoW3

The current version has the same bug as it first had. One point of interest is that if you open and then close and then re-open the dialog box it works first time. That means as far as I know the code has only run the following message blocks:

Code:

switch (message)
{
    case WM_INITDIALOG:
    {
        setAdapterBufferField(text, hDlg); // this works - fills out text fields for adapter
        D3DObj.LogResolution(D3DObj.mdxgiFactory); // this works - fills class vector with display data
        setResolution(hDlg); // this works - loads class data into combo box fields and strings for selection
        SendMessage(GetDlgItem(hDlg, IDC_COMBO1), CB_SETCURSEL, (WPARAM)0, (LPARAM)0); // this works - sets starting selection in combo box
        setRefresh(hDlg); // produces garbage value on first run
        return TRUE; // setting true or false makes no difference (seemingly)
    }

    case WM_COMMAND:
    {
        if (LOWORD(wParam) == IDC_BUTTON1)
        {
            EndDialog(hDlg, LOWORD(wParam));
            return TRUE;
        }
.... switch block continues....

So it's only run the WM_INITDIALOG block and the WM_COMMAND part up to the quit button that ends the dialog box. But re-opening it after this it works perfectly.

Please note this is compiled in Visual Studio and includes a substantial amount of directX stuff to compile. I don't remember exactly how I set it up but I assume one must ensure VS is configured to be able to access all that, otherwise it may not compile. You can still see the program anyway if you navigate to the Debug folder and run the .exe. I'm not which version the Release folder has in it so probably best to ignore that one.

If you have any issues accessing the file please let me know thanks

*p.s. I let someone else look at the code before uploading and they think it might be a message issue as well

Quite a big file? 560MB, expanding to almost 3 gigs, is utter madness!
99.9% of the size comes from the pre-compiled headers (the ".vs" dir) which are only there to speed up your compilation process.
People trying to help do not need that.
Neither do they need the compiled program ("x64" dir containing Debug and Release).
There's some other stuff that's not necessary, but it's small enough that it doesn't matter.
So at most you should have included everything except the ".vs" and "x64" directories.
Here's a 43.6KB zip (< 0.01% of your original upload) that just contains that stuff:
Filebin | 52qpepn3jyrj1wfg
This will only be available for 6 days so you might want to upload it to OneDrive.

BTW, I do not use Windows so I will not be trying to run it.
I'm also not a Windows expert (I hardly remember anything about it since I last used it about 20 years ago).
Hopefully someone who knows more about Windows will take a look, too.

At any rate...

It looks like if I was able to run your program it would show a blank window with a menu.
Choosing "Options" from the menu will start the dialog box.
When the dialog box inits for the first time, it will show garbage as the "Refresh Rate".
If you either choose another resolution from the combo box or exit and reopen the dialog box, the refresh rate will display correctly.
Also, you have the comment "change nI variable to 0 and it works (for first entry)", so it would seem that if you set nI to 0 in the code (instead of trying to retrieve it with ComboBox_GetCurSel) then it works if you just start the program and select "Options" from the main menu.
I also assume that it acts like this in both Debug and Release mode.

Is that right?

Also, my idea with your debug code in setRefresh was to not call SendMessageW (or ComboBox_GetCurSel) twice since that's not what your non-debug code does. So try this and let me know the result:

Code:

void setRefresh(HWND hDlg)
{
    WPARAM nI = ComboBox_GetCurSel(GetDlgItem(hDlg, IDC_COMBO1));
    auto _resData = (D3DClass::resData*)SendMessageW(GetDlgItem(hDlg, IDC_COMBO1), CB_GETITEMDATA, nI, 0);
 
    #ifdef SKIP  // change to _DEBUG to display
    std::wstring _string = std::to_wstring(_resData->w)  + L" "
                         + std::to_wstring(_resData->h)  + L"\n"
                         + std::to_wstring(_resData->Hz) + L"\n"
                         + std::to_wstring(nI)           + L"\n\n"; // may as well print nI, too
    OutputDebugString(_string.c_str());
    #endif
 
    D3DObj.mRefreshRate = _resData->Hz;
    SetWindowTextW(GetDlgItem(hDlg, IDC_EDIT3), std::to_wstring(_resData->Hz).c_str());
}

Hi there,

I've removed the original and uploaded a small version based on your advice John. Now I know what to do (and what not to do) in the future

https://1drv.ms/u/s!AmIumtq4Ent9hje6..._aaso?e=4PGKPF

Your diagnosis is correct, with the only exception being that user made selection changes in the combo box on the first run make no difference to the bug.

If you check the check box and re-select an entry in the combo box, or if you simply close the dialog box and re-open it, it works.

I've made the code changes you prescribe in the new upload also (although the bug persists).

Thanks for the advice John.

*p.s. also you are correct about forcing the index to zero instead of using the WPARAM variable nI. It will successfully fill out just the first entry correctly with no garbage value in the Refresh Rate field.

I've made the code changes you prescribe in the new upload also

Note that I made a small change in the setRefresh debug code to print nI as well.
What does that debug code print on the first opening of the dialog box?

Hi there,

Okay it's working. But I don't really know exactly what I did. I wrote a small bit of code to display the messages the dlgproc was receiving and I found a list online that tells you which messages the numbers correspond to.

It seems the dlgproc receives a WM_SETFONT message and then after that an WM_INITDIALOG message, then shortly after that it receives a WM_ACTIVATE message.

Here's the code to retrieve them right at the start of the dlgproc:

Code:

#ifdef _DEBUG  // change to _DEBUG to display
    std::wstring _string = L"Message: " + std::to_wstring(message) + L"\n\n";
    OutputDebugString(_string.c_str());
#endif

And the debug outputs (at the top):



and further down:



So I decided to try just setting the adapter and format text fields in WM_INITDIALOG and leave the rest to the WM_ACTIVATE block. I felt maybe the activate message was a bit more "lively" for some reason.

When I removed the return from the WM_INITDIALOG block it looks like this:

Code:

LRESULT CALLBACK ChooseRender(HWND hDlg, UINT message, WPARAM wParam, LPARAM lParam)
{
    WCHAR* text = nullptr;
    LRESULT checkBox = Button_GetCheck(GetDlgItem(hDlg, IDC_CHECK1));

    #ifdef _DEBUG  // change to _DEBUG to display
        std::wstring _string = L"Message: " + std::to_wstring(message) + L"\n\n";
        OutputDebugString(_string.c_str());
    #endif

    switch (message)
    {       
        case WM_INITDIALOG:
        {
            setAdapterBufferField(text, hDlg); // this works - fills out text fields for adapter
        }

        case WM_ACTIVATE:
        {
            D3DObj.LogResolution(D3DObj.mdxgiFactory); // this works - fills class vector with display data
            SendMessageW(GetDlgItem(hDlg, IDC_COMBO1), CB_RESETCONTENT, 0, 0); // clear combo box
            setResolution(hDlg); // this works - loads class data into combo box fields and strings for selection
            SendMessage(GetDlgItem(hDlg, IDC_COMBO1), CB_SETCURSEL, (WPARAM)0, (LPARAM)0); // this works - sets starting selection in combo box
            setRefresh(hDlg); // this now also works
            return TRUE;
        }

.... dlgproc continues

And that works fine. Beyond a hunch that was successful......I have absolutely no idea why this is working, but now I've found a solution perhaps it will be easier for others to tell me why.

I don't want to just leave it at that I feel I really need to understand why this is happening. Anyways I'm pleased it's working though.

Thanks very much for the help John and anyone else who may post future replies

The code above does the following for WM_INITDIALOG:

Code:

setAdapterBufferField(text, hDlg);
D3DObj.LogResolution(D3DObj.mdxgiFactory);
SendMessageW(GetDlgItem(hDlg, IDC_COMBO1), CB_RESETCONTENT, 0, 0);
setResolution(hDlg);
SendMessage(GetDlgItem(hDlg, IDC_COMBO1), CB_SETCURSEL, (WPARAM)0, (LPARAM)0);
setRefresh(hDlg);
return TRUE;

And then the following (repeating most of the above) for WM_ACTIVATE:

Code:

D3DObj.LogResolution(D3DObj.mdxgiFactory);
SendMessageW(GetDlgItem(hDlg, IDC_COMBO1), CB_RESETCONTENT, 0, 0);
setResolution(hDlg);
SendMessage(GetDlgItem(hDlg, IDC_COMBO1), CB_SETCURSEL, (WPARAM)0, (LPARAM)0);
setRefresh(hDlg);
return TRUE;

How do you know it's not just repeating the code that is making it work?
To test it, get rid of the WM_ACTIVATE handler and try:

Code:

case WM_INITDIALOG:
  setAdapterBufferField(text, hDlg);
  D3DObj.LogResolution(D3DObj.mdxgiFactory);
  SendMessageW(GetDlgItem(hDlg, IDC_COMBO1), CB_RESETCONTENT, 0, 0);
  setResolution(hDlg);
  SendMessage(GetDlgItem(hDlg, IDC_COMBO1), CB_SETCURSEL, (WPARAM)0, (LPARAM)0);
  setRefresh(hDlg);
  D3DObj.LogResolution(D3DObj.mdxgiFactory);
  SendMessageW(GetDlgItem(hDlg, IDC_COMBO1), CB_RESETCONTENT, 0, 0);
  setResolution(hDlg);
  SendMessage(GetDlgItem(hDlg, IDC_COMBO1), CB_SETCURSEL, (WPARAM)0, (LPARAM)0);
  setRefresh(hDlg);
  return TRUE;

I'm not saying that's reasonable, but if it also works then it is because of some of the repetition and has nothing to do with WM_ACTIVATE.