Re: [4/5] GetWindowLongPtr: Windows

[Dmitry Timoshkov]

"Robert Shearman"  wrote:
> - if (!(context.Eax = GetWindowWord( HWND_32(hwnd), GWL_HINSTANCE ))) context.Eax = context.SegDs;
> - if (!(context.Eax = GetWindowWord( HWND_32(hwnd), GWLP_HINSTANCE ))) context.Eax = context.SegDs;

GetWindowWord above probably should be replaced by GetWindowLongPtrW.
And all appropriate ANSI variants of the GetWindowLongPtr call should
have been replaced by a unicode one while you were at it.

Of course that's not critical at all, just a nit pick.

[4/5] GetWindowLongPtr: Windows

[Robert Shearman]

Dmitry Timoshkov wrote:
> "Robert Shearman"  wrote:
> > - if (!(context.Eax = GetWindowWord( HWND_32(hwnd), GWL_HINSTANCE ))) context.Eax = context.SegDs;
> > - if (!(context.Eax = GetWindowWord( HWND_32(hwnd), GWLP_HINSTANCE ))) context.Eax = context.SegDs;
>
> GetWindowWord above probably should be replaced by GetWindowLongPtrW.
> And all appropriate ANSI variants of the GetWindowLongPtr call should
> have been replaced by a unicode one while you were at it.
>
> Of course that's not critical at all, just a nit pick.

Yes, but at least GetWindowWord warns when discarding high bytes of the 
instance handle. Besides, this code probably won't compile on a 64-bit 
build anyway.
I did replace some GetWindowLongA calls with GetWindowLongPtrW, but 
there is really no difference between the two A and W versions in terms 
of performance anyway.

Rob

[4/5] GetWindowLongPtr: Windows

[Dimitrie O. Paun]

On Thu, Sep 23, 2004 at 01:37:32PM +0100, Robert Shearman wrote:
> I did replace some GetWindowLongA calls with GetWindowLongPtrW, but 
> there is really no difference between the two A and W versions in terms 
> of performance anyway.

Yes, but we should use the W version in preference to the A one as a
matter of principle. This way, a simple syntactic check can tell us
that we don't have W->A transitions. Otherwise, we have to do semantic
checks all the time, and that's just not fun :)