question regarding dibsections

[Marcus Meissner]

Hi,

While debugging twain, I came across the nativexfer method, used by 60-70%
of all twain using programs.

It has some very unfortunate programming logic.

Here is a sample snippet from:
    http://hp.vector.co.jp/authors/VA011973/prg_twain.htm

TW_UINT16 rc;
    TW_UINT32 hBitmap;
    TW_BOOL PendingXfers = TRUE;
    LPBITMAPINFOHEADER      lpDib;

hBitmap = NULL;
    rc = lpfnDSM_Entry(pAppId,pSourceId,DG_IMAGE,DAT_IMAGENATIVEXFER,MSG_GET,(TW_MEMREF)&hBitmap);
    lpDib = (LPBITMAPINFOHEADER)GlobalLock((void*)hBitmap);
    /* ... */
    if(lpdib!=NULL){
    	GlobalUnlock((void*)hBitmap);
    	GlobalFree((void*)hBitmap);
    }

(hBitmap is the return value of CreateDIBSection()).

This sample code apparently assumes that GlobalLock(HBITMAP) gives it
a pointer to a BITMAPINFOHEADER.

We do not do this, for us HBITMAP is a regular 16bit GDI
handle.  GlobalLock will not work on it :/

Any ideas?

Ciao, Marcus

[Dimi Paun]

On Mon, April 24, 2006 3:44 pm, Marcus Meissner said:
> We do not do this, for us HBITMAP is a regular 16bit GDI
> handle.  GlobalLock will not work on it :/

Not sure if its related, but this thread:
    http://www.winehq.org/pipermail/wine-devel/2006-April/thread.html#46804

In particular Alexandre's comment:
    http://www.winehq.org/pipermail/wine-devel/2006-April/046819.html
on handling the 16-bit case in a specific manner.

We may have to do that here as well.

HTH.

[Marcus Meissner]

On Mon, Apr 24, 2006 at 09:44:50PM +0200, Marcus Meissner wrote:
> Hi,
>
> While debugging twain, I came across the nativexfer method, used by 60-70%
> of all twain using programs.
>
> It has some very unfortunate programming logic.
>
> Here is a sample snippet from:
>    http://hp.vector.co.jp/authors/VA011973/prg_twain.htm
>
> TW_UINT16 rc;
>    TW_UINT32 hBitmap;
>    TW_BOOL PendingXfers = TRUE;
>    LPBITMAPINFOHEADER      lpDib;
>
> hBitmap = NULL;
>    rc = lpfnDSM_Entry(pAppId,pSourceId,DG_IMAGE,DAT_IMAGENATIVEXFER,MSG_GET,(TW_MEMREF)&hBitmap);
>    lpDib = (LPBITMAPINFOHEADER)GlobalLock((void*)hBitmap);
>    /* ... */
>    if(lpdib!=NULL){
>    	GlobalUnlock((void*)hBitmap);
>    	GlobalFree((void*)hBitmap);
>    }
>
> (hBitmap is the return value of CreateDIBSection()).
>
> This sample code apparently assumes that GlobalLock(HBITMAP) gives it
> a pointer to a BITMAPINFOHEADER.

It also expects the DIB bits right behind this header (+ palette), which absolutely
does not match our currently DIB memory layouts.

I expect much fun :)

Ciao, Marcus