| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
| |
Make sure the plugin builds but doesn't do anything when configured
without -plugin-gfx-directfb
Reviewed-by: Donald Carr <donald.carr@nokia.com>
|
|
|
|
| |
Reviewed-by: Donald Carr <donald.carr@nokia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Since Qt 4.5, all tablet events are compressed not to overload the
widgets with the corresponding mouse event (a mouse event is generated
if the tablet event is not accepted).
This behavior reduce the precision when drawing on a widget that use
the tablet events. All tablet events should be sent to the widget that
are accepting the tablet event.
With this patch, the tablet event are filtered only if the widget
ignore the first tablet event. The mouse events are compressed.
There is two special cases for the filtering:
First, if a tablet event is for another widget than the one
ignoring the tablet, this event should not be filtered.
Second, if there is a mouse press event, the mouse move event should be
sent to the widget that received the mouse press event.
Helped-by: Pierre Rossi
Reviewed-by: Thomas Zander
Reviewed-by: Bradley T. Hughes
|
|
|
|
|
| |
For the first tablet events, polyLine contains null points. The first
two line painted have an invalid origin.
|
|
|
|
|
|
| |
Make sure all files are wrapped in QT_NO_QWS_DIRECTFB
Reviewed-by: Donald Carr <donald.carr@nokia.com>
|
|
|
|
| |
Reviewed-by: Donald Carr <donald.carr@nokia.com>
|
|
|
|
| |
Reviewed-by: Donald Carr <donald.carr@nokia.com>
|
|
|
|
| |
Reviewed-by: Donald Carr <donald.carr@nokia.com>
|
|
|
|
| |
Reviewed-by: Donald Carr <donald.carr@nokia.com>
|
|
|
|
|
|
|
|
|
| |
That method can be slow, since it does a DHCP query followed by a DNS query
(sometimes followed by a Netbios query, too)... no point in doing it for every
file URL loaded in webkit.
Merge-request: 1128
Reviewed-by: Thiago Macieira <thiago.macieira@nokia.com>
|
|
|
|
|
|
|
|
|
| |
From libpng changelog:
version 1.4.0beta74 [August 8, 2009]
Changed png_ptr and info_ptr member "trans" to "trans_alpha".
Merge-request: 1317
Reviewed-by: Thiago Macieira <thiago.macieira@nokia.com>
|
|
|
|
|
|
|
|
| |
Qt expects a trans_values member in png_info_struct; this member has
been renamed to trans_color in libpng 1.4.
Merge-request: 1317
Reviewed-by: Thiago Macieira <thiago.macieira@nokia.com>
|
|
|
|
|
|
|
| |
Signed-off-by: Riccardo Iaconelli <riccardo@kde.org>
Merge-request: 1371
Reviewed-by: Thiago Macieira <thiago.macieira@nokia.com>
|
| |
|
| |
|
|
|
|
|
|
|
| |
Simply following the system configuration for the proxy used to require
writing a QNetworkProxyFactory subclass. The static setter makes this
easier, so apps can in one line say "I want to use the system proxy
settings". Solution and method name suggested by Thiago.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Since we need QMargins for other things then
the CSS helper functions in drawutil, we have to make it more
generic. It is already useful for QWidget::contentsMargins
for example. This ensures we have some flexibility on how to
use and modify it in the future.
Reviewed-by: mbm
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Consolidated Interlocked* declarations and API implementation through
macro hackery, (hopefully) for improved readability and maintainability.
Fixes anti-aliasing warnings with MinGW in qatomic_windows.h. Gcc
builds now use inline assembly for atomic operations, instead of relying
on Interlocked* functions which aren't consistently declared across
implementations (mingw32, mingw-w64, wine... others?).
Drops support for VC 6 and MetroWerks.
Reviewed-by: Thiago Macieira
|
|
|
|
| |
Reviewed-by: Trond
|
|
|
|
| |
Reviewed-by: Trust Me
|
|
|
|
|
|
|
|
|
|
|
|
| |
I renamed QGLContextGroupResources to QGLContextGroup because we are
using it to identify context groups. I also added a pointer to one of
the contexts in the group. Together with qgl_share_reg(), the pointer
can be used to find all contexts in a group.
I renamed QGLContextPrivate::qt_get_extension_funcs() to
QGLContextPrivate::extensionFuncs() to follow Qt's naming convention.
Reviewed-by: Trond
|
|
|
|
|
|
|
| |
Create a specialized version of numbers parsing that works on a short
QVarLengthArray since a transformation matrix has at most 6 elements.
Reviewed-by: Kim
|
|
|
|
|
|
|
|
|
|
|
|
| |
Instead of comparing the character to '0' and '9', use bit fiddling
to detect that the character is a digit between '0' and '9'.
Loading tiger.svg (tests/benchmarks/qsvgrenderer) is now 10% faster,
going down from 85.3 millions instructions to 77.2 millions.
Mostly this is due 46% speed-up in parseNumbersList() function,
from 26.9 millions instructions to just 18.4 millions.
Reviewed-by: Kim
|
|
|
|
|
|
|
|
| |
This reverts a change added by 6a3de1f5 by mistake (according to the
S60 guys). The proper fix may be to add a -small-screen argument,
or to run fullscreen on embedded only.
Reviewed-by: jbarron
|
|
|
|
|
|
| |
Oops, forgot to remove the autotest after rewriting the gestures api.
Reviewed-by: trustme
|
|
|
|
|
|
|
| |
Blending should not be enabled when copying the font cache texture
into the fbo. It *may* cause artifacts with some drivers.
Reviewed-by: Samuel
|
|
|
|
| |
The data is essentially random line noise that happens to live in a .cpp file.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If you show a child panel of an active panel, the child will now be
activated and the parent deactivated. Hiding the child panel will
reactive the parent. If the parent is 0, no other panel is activated.
Reparenting a panel onto an active panel will also activate the (new)
child. Reparenting away does not affect activation in any way.
This change also fixes QGraphicsWidget::isActiveWindow(), which returned
true for all toplevel widgets (not in a panel/window). This is wrong;
either the non-panel items are active, or a panel is active. The
correct behavior is the same as calling QGraphicsItem::isActive(). Fixed
the autotests (which wrongly tested that both a panel and a non-panel
item were active at the same time).
This change causes popups (QGraphics{Proxy,}Widget) to deactivate the
parent widget. On the positive side this activates the popup, and ensures
that the parent regains proper focus when the popup is closed. However
it also means the parent widget is inactive while the popup is open,
which (e.g.) causes editable combobox line edit cursors to stop blinking.
This is to be fixed soon, but the fix is a bit big so we'll do that later.
Autotests included.
Reviewed-by: Brad
|
|
|
|
|
|
|
|
|
| |
Allow delayed activation for more fine grained control over which panels
are activated or left inactive when the scene is created.
Autotests included.
Reviewed-by: Brad
|
|
|
|
| |
Reviewed-by: Kim
|
|
|
|
|
|
|
|
| |
Instead of doing an attribute look-up via QXmlAttributes::value(),
we just iterate by ourselves. Thus, we need to carry out the iteration
and comparison only once, instead of every call to the said value().
Reviewed-by: Kim
|
|
|
|
|
|
|
|
|
|
|
| |
Made the GL blur filter code slightly less hacky by not reusing the same
paint engine for rendering both to the offscreen FBO and to the actual
target device. This should make the code less reliant on paint engine
implementation details and thus more robust with regards to changes in
the paint engine.
Task-number: 260402
Reviewed-by: Trond
|
|
|
|
|
|
|
| |
It's insufficient to use a single paint engine to render to all FBOs. If
the default engine is already in used we need to create our own engine.
Reviewed-by: Trond
|
|
|
|
|
| |
We should use the newly create QStringRef, after all that is the idea
of the optimization.
|
| |
|
|\ |
|
| |
| |
| |
| | |
Reviewed-by: Kim
|
| |
| |
| |
| |
| |
| | |
Use QStringRef operations as much as possible.
Reviewed-by: Kim
|
| |
| |
| |
| |
| |
| | |
Compiler error: hex escape sequence out of range.
Reviewed-by: ogoffart
|
| |
| |
| |
| |
| |
| |
| | |
Oftentimes and depending on your window manager, QTest::mouseClick is not
enough to set the focus on a widget. QTest::mouseMove to that widget first.
Reviewed-by: ogoffart
|
| |
| |
| |
| | |
Reviewed-by: Kim
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Shader objects had a pointer to the context they were originally
created in. If the context was destroyed, the shader would (on Windows)
dereference an invalid pointer and cause the program to crash. I
replaced the context pointer with a pointer to the context group. I
also added checks in debug mode to make sure the context associated
with the shader shares resources with the current context.
Reviewed-by: Tom
|
| |
| |
| |
| | |
Reviewed-by: Kim
|
| |
| |
| |
| | |
Reviewed-by: Kim
|
| |
| |
| |
| | |
Reviewed-by: Kim
|
| |\ |
|
| | |
| | |
| | |
| | |
| | |
| | | |
Including the befaviour change
Reviewed-by: Thierry
|
| |/
| |
| |
| | |
Reviewed-by: Trust Me
|