COLORONCOLOR(STRETCH_DELETESCANS) was used in place of HALFTONE. COLORONCOLOR mode may delete rows
of pixels without trying to preserve information so it will cause Wine to render poorly when the
destination rectangle is small.
According to tests, HALFTONE mode uses box filter when doing integer downscaling and nearest
neighbor interpolation when doing upscaling in both horizontally and vertically. In other cases,
HALFTONE mode uses a lanczos3 like algorithm to interpolate pixels. There are also other heuristics
involved. For example, shrinking a 2x2 image to 1x1 may not use box filter depending on image data.
Since this algorithm is undocumented, it's difficult to reverse engineer the original algorithm and
produce identical results. Instead, this patch uses a naive implementation of bilinear interpolation
to implement HALFTONE mode and it produces good quality images.
For 8-bit and lower color depth images, nulldrv_StretchBlt should resize the images first and then
converts color depth.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=46375
Signed-off-by: Zhiyi Zhang <zzhang@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
In particular this stops the traces in open_clipboard() and
has_no_open_wnd() from clobbering the line number when used in
expressions such as ok(has_no_open_wnd(), ...).
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
One of the tests expects GetLastError() to still return 0xdeadbeef after
has_no_open_wnd(), which would not be the case if another process did
open the clipboard.
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
As for OpenClipboard(), if another application opened the clipboard,
GetOpenClipboardWindow() will not return the expected value.
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Applications using ole32's clipboard API (e.g. Radeon / QT5) end up
monitoring the content of the clipboard and calling OpenClipboard()
while user32:clipboard runs. This causes the test's own
OpenClipboard() calls to fail.
Similarly the KDE clipboard manager may query the Wine clipboard
content while the test runs, causing winex11.drv to call
OpenClipboard() with the same effect.
So call OpenClipboard() again when it fails due to the clipboard
already being open. If opening the clipboard still fails, trace who
opened the clipboard, particularly the window class name to help
identify the source of interference.
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Buttons are really just values with BitSize == 1, and having a union for
that makes the code more complicated than it needs.
We'll simplify the code a lot using HIDP_VALUE_CAPS everywhere, as the
two structure types are compatible.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
This ensures all traces have the relevant context information, in
particular the OpenClipboard() checks, and simplifies the ok() calls.
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Without any HID report data for now.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=50506
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Otherwise both on Linux and Mac, the window will not restore title bar
when it becomes non-shaped until it's e.g. moved.
Signed-off-by: Piotr Caban <piotr@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
One cannot assume that a random integer has zero chance of being a valid
atom.
Reenable the GlobalFindAtomA() check in a todo_wine.
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
By default CF_LOCALE matches the current input language, not the default
user LCID.
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Which creates an off-screen window surface for top-level non-layered or
SLWA-layered windows.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
This saves calls to GetMonitorInfo() and return from EnumDisplayMonitors() immediately
after the primary monitor is found.
Signed-off-by: Zhiyi Zhang <zzhang@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
When high word is 0, and unless it has an explicit layout id.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
This reduce the time to run the monitor test from 6 minutes to 25 seconds.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=50086
Signed-off-by: Zhiyi Zhang <zzhang@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
So that Wine doesn't have to send an extra WM_NULL message when
wait_graphics_driver_ready() is later called in
EnumDisplayDevicesW(). Also, it avoids a deadlock when
wait_graphics_driver_ready() is used in later patches.
Signed-off-by: Zhiyi Zhang <zzhang@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Zhiyi Zhang