Re: callback problem

Hi,
some time ago I also tried to use Native Function call implemented in JAWIN
but failed. This piece of the project have not been accomplished. They
suggest to use explicit marshaling that is not clear for Developer. Function
calling conventions cdecl, fastcall are not supported (only stdcall).
Structure marshal is headache (they do not support union, struct. packing,
etc.). The same problems with call of native functions from virtual tables.
I also tried to use another projects like xFunctions, JNIWrappers, etc.
These projects support complex explicit marshaling makes my code huge and
unclear.
As a result I started to develop my own API for Native Function Call from
Java. Having learnt Platform Invoke in .NET I have realized that the same
can be done in SUN/IBM Java. My API implements in Java .NET Platform Invoke
paradigm:
- implicit and explicit marshaling;
- C Structure marshaling;
- Java array marshaling;
- marshal by value and by reference;
- callback functions;
- C class functions call from Class Virtual Table, etc.

See links:
http://www.simtel.net/product.php[id]100916[SiteID]simtel.net
http://www.programmersheaven.com/download/51865/download.aspx



Regards,
Vitaly Shelest,


"Howard Chivers" <[email protected]> wrote in message
news:LISTSERV%[email protected]...
> Hi,
>
> I've been trying to use jawin to wrap some User32 functions - many thanks
> for the framework & I'll be happy to donate any stubs that result.
> However, I'm having problems with callbacks.
>
> I'd be grateful if anyone could have a quick look at the following; I
> think
> I've read the source correctly & it seems straightforward, but I'd be
> grateful for confirmation.
>
> The following example wraps the EnumWindows function. Similar problems
> occur with other enum functions. The code works correctly, and calls back
> to EnumWindowsProc, passing handles and lparam correctly; after exactly 5
> callbacks (and with lots more to enumerate) the system throws a protection
> exception - either protected code or access violation. (A more complex
> example with enum child has convinced me that this is not a genuine
> security problem).
>
> It is possible that I have not correctly worked out how to build the
> return
> from the callback, with the result that the stack is mangled on return
> from
> callback? or is this likley to be a bug?
>
> Thanks
> Howard
>
> --------
> public void debugEnumWindows() throws COMException, IOException {
>        User32_a.EnumWindows(new EnumWindowsProc() {
>            public boolean call(int hWnd,int lParam) {
>                return true;
>                }
>           },0x99);
>        return;
>    }
> --------------
> public static void EnumWindows(EnumWindowsProc ewp,int lParam) throws
> IOException, COMException {
>        FuncPtr fp = new FuncPtr("user32.dll","EnumWindows");
>        NakedByteStream nbs = new NakedByteStream();
>        LittleEndianOutputStream leos = new LittleEndianOutputStream
> (nbs);
>        leos.writeInt(0);
>        leos.writeInt(lParam);
>        byte[] res = fp.invoke("T2." + EnumWindowsProc.mtoken + "I::",
>                8, nbs, new Object[] {ewp}, ReturnFlags.CHECK_FALSE);
>        return;
>    }
> -----------------
>
> public abstract class EnumWindowsProc extends WinFuncPtr {
>
>    public static final int mtoken = GenericStub.registerCustomString
> ("2II:I:");
>    abstract public boolean call (int hWnd,int lParam) throws IOException;
>
>    protected byte[] call(byte[] inargs, Object[] inobjs) throws
> IOException {
>        LittleEndianInputStream leis = new LittleEndianInputStream(new
> ByteArrayInputStream(inargs));
>        int hWnd = leis.readInt();
>        int lParam = leis.readInt();
>        boolean ret = call(hWnd,lParam);
>        NakedByteStream nbs = new NakedByteStream();
>        LittleEndianOutputStream leos = new LittleEndianOutputStream(nbs);
>        leos.writeInt(ret ? 1 : 0);
>        return nbs.getInternalBuffer();
>    }
> }
>
>
> ----end------
>