Re: user32: Make message test pass cleanly under XP SP2

[Alexandre Julliard]

Dmitry Timoshkov  writes:
>  /* posted message */
>  PostMessageA( hwnd, WM_CHAR, dbch[0], 0 );
> -
>
> msg.message = 0xdeadbeef;
> ok( !PeekMessageW( &msg, hwnd, 0, 0, PM_REMOVE ), "got message %x\n", msg.message );
> PostMessageA( hwnd, WM_CHAR, dbch[1], 0 );
> ok( PeekMessageW( &msg, hwnd, 0, 0, PM_REMOVE ), "no message\n" );
> ok( msg.message == WM_CHAR, "unexpected message %x\n", msg.message );
> ok( msg.wParam == wch, "bad wparam %lx/%x\n", msg.wParam, wch );
> -
>
> msg.message = 0xdeadbeef;
> ok( !PeekMessageW( &msg, hwnd, 0, 0, PM_REMOVE ), "got message %x\n", msg.message );
>
> /* posted thread message */
> PostThreadMessageA( GetCurrentThreadId(), WM_CHAR, dbch[0], 0 );
> -
>
> msg.message = 0xdeadbeef;
> ok( !PeekMessageW( &msg, hwnd, 0, 0, PM_REMOVE ), "got message %x\n", msg.message );
> PostMessageA( hwnd, WM_CHAR, dbch[1], 0 );
> ok( PeekMessageW( &msg, hwnd, 0, 0, PM_REMOVE ), "no message\n" );
> ok( msg.message == WM_CHAR, "unexpected message %x\n", msg.message );
> ok( msg.wParam == wch, "bad wparam %lx/%x\n", msg.wParam, wch );
> -
>
> msg.message = 0xdeadbeef;
> ok( !PeekMessageW( &msg, hwnd, 0, 0, PM_REMOVE ), "got message %x\n", msg.message );

I'm not sure I see the point of that change, since the message is not
tested when PeekMessage fails.

user32: Make message test pass cleanly under XP SP2

[Dmitry Timoshkov]

"Alexandre Julliard"  wrote:
> Dmitry Timoshkov  writes:
> >  /* posted message */
> >  PostMessageA( hwnd, WM_CHAR, dbch[0], 0 );
> > -
> >
> > msg.message = 0xdeadbeef;
> > ok( !PeekMessageW( &msg, hwnd, 0, 0, PM_REMOVE ), "got message %x\n", msg.message );
> > PostMessageA( hwnd, WM_CHAR, dbch[1], 0 );
> > ok( PeekMessageW( &msg, hwnd, 0, 0, PM_REMOVE ), "no message\n" );
> > ok( msg.message == WM_CHAR, "unexpected message %x\n", msg.message );
> > ok( msg.wParam == wch, "bad wparam %lx/%x\n", msg.wParam, wch );
> > -
> >
> > msg.message = 0xdeadbeef;
> > ok( !PeekMessageW( &msg, hwnd, 0, 0, PM_REMOVE ), "got message %x\n", msg.message );
> >
> > /* posted thread message */
> > PostThreadMessageA( GetCurrentThreadId(), WM_CHAR, dbch[0], 0 );
> > -
> >
> > msg.message = 0xdeadbeef;
> > ok( !PeekMessageW( &msg, hwnd, 0, 0, PM_REMOVE ), "got message %x\n", msg.message );
> > PostMessageA( hwnd, WM_CHAR, dbch[1], 0 );
> > ok( PeekMessageW( &msg, hwnd, 0, 0, PM_REMOVE ), "no message\n" );
> > ok( msg.message == WM_CHAR, "unexpected message %x\n", msg.message );
> > ok( msg.wParam == wch, "bad wparam %lx/%x\n", msg.wParam, wch );
> > -
> >
> > msg.message = 0xdeadbeef;
> > ok( !PeekMessageW( &msg, hwnd, 0, 0, PM_REMOVE ), "got message %x\n", msg.message );
>
> I'm not sure I see the point of that change, since the message is not
> tested when PeekMessage fails.

I added that changes because Microsoft compiler from latest Vista SDK
complains that variables might be used without being initialized.

user32: Make message test pass cleanly under XP SP2

[Alexandre Julliard]

"Dmitry Timoshkov"  writes:
> I added that changes because Microsoft compiler from latest Vista SDK
> complains that variables might be used without being initialized.

It sounds like that compiler is on crack. I don't think we want to
sprinkle assignments all over the place just to work around that.