v2:
Add Proper QueryInterface
Fixed Initial Reference Count value.

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=46534Signed-off-by: Esdras Tarsis <esdrastarsis@gmail.com>

Signed-off-by: Alistair Leslie-Hughes <leslie_alistair@hotmail.com>

This shows that the data returned by the EnumKey call isn't the same type
as vbscript.  At a source leve EnumKey returns a VT_BSTR|VT_ARRAY whereas
vbscript converts it a VT_ARRAY|VT_VARIANT*.

This isn't correct and the patch 4 show that this is the case.

The array being returned by Invoke, needs to convert the datatype
on return.
Signed-off-by: Alistair Leslie-Hughes <leslie_alistair@hotmail.com>

Wine-bug: https://bugs.winehq.org/show_bug.cgi?id=51120
Wine-bug: https://bugs.winehq.org/show_bug.cgi?id=39463

This patch supersedes both '[2/2] version: Test for VerQueryValueA'
and '[1/2] version: Fix wrong length reported when selecting a block.
(resend 2)', as it is a better tests, and proves the fix to not work
fully.

The patch shows a difference in handling between wine's VerQueryValue,
and windows' VerQueryValue when selecting a 'BLOCK' object.

Implemented pointer checking and commented the binary structure as
suggested by nsivov from #winehackers.

Testcase by Michael Müller <michael@fds-team.de>.

Windows only sends the activation messages and hooks once, until the
SetActiveWindow completes for that window.
Signed-off-by: Gabriel Ivăncescu <gabrielopcode@gmail.com>

When activating a window and sending activation messages to the window
procedure, Windows avoids a recursive loop by not sending more of these
messages or hooks while it's still activating the window. Some applications
actually depend on this behavior, so it is needed.

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=46274Signed-off-by: Gabriel Ivăncescu <gabrielopcode@gmail.com>

This makes relative raw input independent from cursor speed, as it is
the case on Windows. Absolute raw input is already translated to
virtual desktop space, and cursor speed is meaningless in this case.

This does not support mixed relative/absolute X/Y axis.

Valuator names aren't well specified, and although they usually are
Rel X/Y or Abs X/Y, there are cases where the X/Y names are something
else. Just assume that the first two valuators are the X/Y axes, as it
seems to be generally the case.

For rawinput messages, as user32 is currently only expecting relative
motion.

We still need to send "normal" input from the clipping window thread
to trigger low-level hooks callbacks when clipping cursor. This is for
instance used in our dinput implementation.

So we can generate legacy messages only by calling __wine_send_input
with NULL rawinput, and generate WM_INPUT messages only by calling
__wine_send_input with INPUT_HARDWARE input type and a rawinput.

We are going to receive raw button events and we will need to apply the
correct button mappings ourselves.

Original patch by Andrew Eikum <aeikum@codeweavers.com>.

Under XInput2 protocol version < 2.1, RawEvents are not supposed to be
sent if a pointer grab is active. However slave device events are still
received regardless of this specification and Wine implemented a
workaround to receive RawEvents during pointer grabs by listening to
these slave device events. Then, as soon as a mouse button is pressed
only the grabbing client will receive the raw motion events.

By advertising the support of XInput2 version >= 2.1, where RawEvents
are sent even during pointer grabs, we ensure to receive the RawMotion
events from the desktop window thread, even if a mouse grab is active.

It is now also possible to simplify the code by listening to master
device events only and get rid of slave device id tracking.

This will allow us to listen to the XInput version of several events,
which can bring additional information.

And rename the library and function loader to x11drv_xinput_load.

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=17205Signed-off-by: Alistair Leslie-Hughes <leslie_alistair@hotmail.com>

user32: Fix messages sent on a window without WS_CHILD, but with an invisible parent, when it is shown.

Some applications depend on the fact that WM_WINDOWPOSCHANGING is sent
after WM_SHOWWINDOW when SetParent is called, even if the window ends up
not visible because its parent is not visible, and occurs when the window
itself does not have the WS_CHILD style set.

This also fixes a TODO message sequence, so that Wine matches Windows.

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=40262Signed-off-by: Gabriel Ivăncescu <gabrielopcode@gmail.com>

user32: Before asking a WM to activate a window make sure that the window is in foreground and not minimized.

This patch fixes iconify action using WM's taskbar buttons for Winamp.

And use it as a backend in user32.SetActiveWindow().

For bug #2155.

FIXME:

* It seems like the color key channels are swapped between Windows and Wine.
  This can be tested by replacing the (128, 128, 128) with something else.

* Add tests with WS_CHILD layered Windows. On Windows this is supported for
  >= Win8, but only when a manifest is present.