X11 binding, XGetWindowProperty, and different behaviour for similar code between D and C++ (D fails)

[Simon Gomizelj]

On Fri, 03 Apr 2009 13:33:21 -0400, Simon Gomizelj <simongmzlj at gmail.com>  
wrote:

> On Thu, 02 Apr 2009 09:59:55 -0400, Jarrett Billingsley  
> <jarrett.billingsley at gmail.com> wrote:
>
>> On Fri, Apr 3, 2009 at 4:40 AM, Simon Gomizelj <simongmzlj at gmail.com>  
>> wrote:
>>> On Fri, 03 Apr 2009 04:09:25 -0400, Simon Gomizelj  
>>> <simongmzlj at gmail.com>
>>> wrote:
>>>
>>>> As a personal learning D/learn X11 project, I thought it would be
>>>> interesting to try to port dwm over to D. Building it up nice and  
>>>> slowly,
>>>> right now I'm trying to collect a list of windows and their states  
>>>> a-la
>>>> dwm.c code.
>>>>
>>>> The D code, which is translated from C code, does not work, it  
>>>> returns:
>>>> X Error of failed request:  BadValue (integer parameter out of range  
>>>> for
>>>> operation)
>>>>   Major opcode of failed request:  20 (X_GetProperty)
>>>>   Value in failed request:  0x2
>>>>   Serial number of failed request:  29
>>>>   Current serial number in output stream:  29
>>>>
>>>> The C++ code, which is the D code translated back, does work. The code
>>>> fails on a call to XGetWindowProperty. Debugging each shows that both  
>>>> the
>>>> C++ and D code pass the same atom, display, and window. Setting
>>>> WM_STATE_ELEMENTS < 2 causes the D code to start to segfault while  
>>>> the C++
>>>> code still seems to work.
>>>>
>>>> The std.c.linux.X11.X and std.c.linux.X11.Xlib modules I grabbed off  
>>>> of
>>>> http://www.dsource.org/projects/bindings/browser/trunk/X11?order=date.  
>>>> The
>>>> rest are hand translated (but unconsequential, XGetWindowProperty is
>>>> contained in Xlib). X.d and Xlib.d look correct enough.
>>>>
>>>> I've spent a full day looking at this code, and I can't seem to  
>>>> determine
>>>> why the D code fails. Anyone have any ideas? Thanks in advance.
>>>>
>>>
>>> C++ code:
>>> long getstate(Window window)
>>> {
>>>    static const long WM_STATE_ELEMENTS = 2L;
>>>
>>>    unsigned long nitems;
>>>    unsigned long leftover;
>>>    Atom xa_WM_STATE, actual_type;
>>>    int actual_format;
>>>    int status;
>>>    unsigned char* p = NULL;
>>>
>>>    xa_WM_STATE = XInternAtom(display, "WM_STATE", false);
>>>    cout << "debug: atom " << xa_WM_STATE << endl;
>>>
>>>    cout << "debug: XGetWindowProperty on window " << window << endl;
>>>    status = XGetWindowProperty(display, window,
>>>                  xa_WM_STATE, 0L, WM_STATE_ELEMENTS,
>>>                  false, xa_WM_STATE, &actual_type, &actual_format,
>>>                  &nitems, &leftover, &p);
>>>
>>>    if(status == 0)
>>>    {
>>>        cout << "RETURN:" << ((p != NULL) ? (long)*p : -1) << "  
>>> leftover:" <<
>>> leftover << " nitems:" << nitems << endl;
>>>        XFree(p);
>>>        return (p != NULL) ? (long)*p : -1;
>>>    }
>>>    return -1;
>>> }
>>>
>>> D code:
>>> long getstate(Window window)
>>> {
>>>    static const long WM_STATE_ELEMENTS = 2L;
>>>
>>>    uint nitems;
>>>    uint leftover;
>>>    Atom xa_WM_STATE, actual_type;
>>>    int actual_format;
>>>    int status;
>>>
>>>    ubyte* p = null;
>>>    scope(exit) XFree(p);
>>>
>>>    byte[] name = cast(byte[])"WM_STATE\0";
>>>    xa_WM_STATE = XInternAtom(display, name.ptr, Bool.False);
>>>    debug output(color.green, "debug: atom {}", xa_WM_STATE);
>>>
>>>    debug output(color.cyan, "debug: XGetWindowProperty on window {}",
>>> window);
>>>    status = XGetWindowProperty(display, window,
>>>                  xa_WM_STATE, 0L, WM_STATE_ELEMENTS,
>>>                  Bool.False, xa_WM_STATE, &actual_type, &actual_format,
>>>                  &nitems, &leftover, &p);
>>>    debug output(color.cyan, "debug: XGetWindowProperty, result:{}",  
>>> status);
>>>
>>>    if(status == 0)
>>>        return (p != null) ? cast(long)*p : -1;
>>>    return -1;
>>> }
>>
>> My best guess is that you've translated those X11 function headers
>> incorrectly.  Perhaps you've used a D long where you shouldn't have.
>> A D long is 64 bits, while a C long is 32, at least on 32-bit
>> platforms.  If you were to pass a 64-bit value where it was expecting
>> a 32-bit one, the params would be in all the wrong places.  But this
>> is just speculation, since I don't know what your bindings look like.
>
> Yeah, thats it exactly, which is annoying because since I downloaded  
> this binding I just assumed it was done right 8-).
>
> extern (C):
> extern int XGetWindowProperty(
>      Display*    /* display */,
>      Window      /* w */,
>      Atom        /* property */,
>      long        /* long_offset */,  <-- should be int
>      long        /* long_length */,  <-- should be int
>      Bool        /* delete */,
>      Atom        /* req_type */,
>      Atom*       /* actual_type_return */,
>      int*        /* actual_format_return */,
>      uint*       /* nitems_return */,
>      uint*       /* bytes_after_return */,
>      ubyte**     /* prop_return */
> );
>
> Since their are no occurences of long longs I just did a search and  
> replace across the whole file for longs to ints
>

For curiousities sake, is there any interest to adding my bindings to any  
standard library? I know bindings for X.h and Xlib.h exist, but I created  
a more complete set: X.d, Xlib.d, Xatom.d, Xproto.d, Xregion.d Xutil.d and  
Xinerama.d and xf86vmode.h

-- 
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/