summaryrefslogtreecommitdiffstats
path: root/src/gui/kernel/qwidget_s60.cpp
Commit message (Collapse)AuthorAgeFilesLines
* Fixed defect in handling of expose events for SymbianGareth Stockwell2010-02-091-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | Commit bc82db did not correctly handle native child widgets. Consider the case when we have a top-level widget A with a native child widget B. When QSymbianControl::Draw() is called on the control corresponding to B, the following occurs: 1. The inExpose flag is set in B's QWExtra structure. 2. The call to syncBackingStore() results in a call to QWidgetBackingStore::flush(), passing default parameters. 3. Because no target widget was passed to flush(), this function selects the top-level widget (A) as the target for the flush operation, passing A as the first argument of QS60WindowSurface::flush(). 4. QS60WindowSurface::flush() checks the inExpose flag from A's QWExtra structure, finds it to be false, and proceeds to call DrawNow() on A's control. Because QSymbianControl::Draw() uses the default graphics context, this context is shared between controls. This means that the DrawNow() call in step 4 causes a WSERV-10 panic (Activate() called on an already-active) graphics context. This patch moves the inExpose flag from B's QWExtra into A's QTLWExtra, with the result that the call to DrawNow() in step 4 is suppressed. Task-number: QTBUG-7960 Reviewed-by: axis
* Whitespace/tab fixes.Jani Hautakangas2010-02-051-1/+1
| | | | Reviewed-by: Trust me
* s60 application loses normalGeometry when returning from fullscreenSamuel Nevala2010-02-051-78/+30
| | | | | | | | | | | | | | | | | | | | | | | Problem description: normalGeomerty lost during showFullScreen 1. Reported problen was due on void QSymbianControl::PositionChanged() over write top->normaGeometry on every position change. As fix top->normalGeometry is moved to new rect:s top left only when widget windowState == 0. 2. Also made some new qwidget auto tests. Refactored s60 side setWindowState to be more readable. Minimized window state now hides window decoration. QApplication & QWidget autotest run on emulator and tested on s60 5.0 hw using attached application. http://bugreports.qt.nokia.com/browse/QTBUG-6231 Task-number:QTBUG-6231 Merge-request: 2256 Reviewed-by: Jani Hautakangas <ext-jani.hautakangas@nokia.com>
* Flush the WSERV command buffer after deleting a surface.Jason Barron2010-02-041-0/+4
| | | | | | | | | | | | | | For graphics systems that use EGL surfaces in the backing store destroying the surface does not guarantee that the memory is immediately freed because this command does not cause a flush. This implies that a manual flush is instead needed. We do this in 2 places; the first is when the surface is destroyed due to a visibility changed. The second case is just after the window has been destroyed. At this point the backing store has already been deleted so the deletion of both the surface and window can happen atomically in WSERV. Task-number: QT-2506 Reviewed-by: Iain
* Enable surface transparency support on Symbian^4.Jason Barron2010-01-141-0/+8
| | | | | | | | | On Symbian^4 systems where the window supports surface transparency, we use this for the Qt::WA_TranslucentBackground flag instead of the previous method. Task-number: QT-2026 Reviewed-by: Iain
* Improve the behavior of expose events on Symbian.Jason Barron2010-01-141-0/+1
| | | | | | | | | | | | | | | | | | | | | | | | Previously when an expose was received from WSERV, we simply called BitBlt (for raster) or called flush on the window surface (for anything else). This behavior differs from other platforms which call syncBackingStore(). This difference means that we flush the backing store without actually updating the content first. This works for most cases because if there actually was new content, it would be updated when the widget's UpdateRequest event was handled. The problem arises when the backing store does not have the correct content. This can happen if the backing store was deleted, but only partially restored (see Task below). Another problem is with the OpenVG graphics system which assumes that beginPaint() is called before endPaint() is order to initialize the context and the surface size. The fix is to call syncBackingStore() like the other platforms, but introduce a bit field to prevent infinite recursion in the painting pipeline. Task-number: QTBUG-4921 Reviewed-by: axis Reviewed-by: Gareth Stockwell
* Prevent a crash when creating an inputContext from the QApplication dtor.Simon Hausmann2010-01-121-1/+1
| | | | | | | | | Complete the fix in f72165460d27860cabd51691f4d935fd74b50f80 by applying the same fix to Symbian and QWS. Task-number: QTBUG-7105 Reviewed-by: Alexis Reviewed-by: Jason McDonald
* Update copyright year to 2010Jason McDonald2010-01-061-1/+1
| | | | Reviewed-by: Trust Me
* Symbian control invokes slots before and after native draw opsGareth Stockwell2009-11-261-0/+1
| | | | | | | | | | | | | | | | | Direct Screen Access (DSA) allows a client to request notification from the window server when drawing is performed by other threads, into a specified region of the screen. This allows DSA rendering - for example video - to be suspended when notifications are drawn, preventing the video content from overwriting the notification. If the drawing originates from the same thread as that which holds the DSA session, DSA must be suspended while drawing takes place. This change allows a widget to request notification when native drawing is about to be performed by QSymbianControl::Draw. Task-number: QTBUG-5467 Reviewed-by: Jason Barron
* Allow Symbian widget implementations to select native paint modeGareth Stockwell2009-11-261-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | On the Symbian platform, the Qt raster paint engine targets an off-screen buffer owned by the Font & Bitmap server (FBSERV). When an area of the screen needs to be refreshed, the window server (WSERV) asks the control environment (CONE) to redraw the control(s) intersecting that screen region. Each Qt native widget has an associated Symbian control, whose Draw function blits the required region of the backing store via WSERV. Use cases involving Direct Screen Access (DSA) may require this behaviour to be modified, to either of the following: - Disable: the Draw function does nothing. In this case, the output of paint events, rendered to the backing store, is not blitted to the screen. This mode was introduced by change 8f445e13. - Zero fill: the Draw function fills all pixels within the redraw region with zeroes. This change allows the widget implementation to select either of these alternative modes by setting a flag in its QWExtra structure. Note that these alternative modes are only suitable for native widgets, because they act on a per-control rather than per-widget basis. Task-number: QTBUG-5467 Reviewed-by: Jason Barron
* Removed window activation hack, unified and fixed title&icon setting.Janne Anttila2009-11-231-18/+0
| | | | | | | | | | | | | | | | | | | | | | | | The window activation hack is not needed anynmore since AppUi()->RemoveFromStack() ensures the next visible window will get the keyboard focus. Hack removal also required change to window title setting logic. Since window title (and icon) are associated to top-level windows, the logical place to set them is the same place where active window is changed i.e. QApplication::setActiveWindow is called. At the same time also window icon setting from show_sys was move to focusChanged to make icon/title setting more consistent. When changing orientation or switching to different statuspane mode we receive KInternalStatusPaneChange events for each window in QSymbianControl::HandleResourceChange. When receiving such event we only need to reset the icon for focused/visible window. If we don't handle it like this, it might happen that invisible widget added to control stack resets the global icon/title. Task-number: QTBUG-5780 Reviewed-by: Axis
* Merge branch '4.6' of git@scm.dev.nokia.troll.no:qt/qt-s60-public into ↵axis2009-11-131-7/+9
|\ | | | | | | 4.6-staging2
| * Fixed maximized dialogs losing maximized status on orientation switchMiikka Heikkinen2009-11-131-7/+9
| | | | | | | | | | | | | | | | | | | | | | | | If maximized dialog had minimum size that didn't fully fit to the screen, it lost its maximized status when orientation was switched, because of the logic that assumed that if a window is resized, it must no longer be maximized. Skipped this assumption for cases where resize occurs because enforcement of the minimum size of the window. Task-number: QTBUG-4671 Reviewed-by: Janne Anttila Reviewed-by: Sami Merila
* | Merge commit upstream/4.6 into oslo-staging-2/4.6Olivier Goffart2009-11-131-1/+1
|\ \ | |/ |/|
| * Fix typo in license headerAlessandro Portale2009-11-121-1/+1
| | | | | | | | | | | | The word 'module' was missing. Reviewed-By: TrustMe
* | Fix for assertion failure in in QWidget::grabMouse for Symbian.Janne Anttila2009-11-111-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | Docs say: "Note that only visible widgets can grab mouse input. If isVisible() returns false for a widget, that widget cannot call grabMouse()." qwidget_x11.cpp uses the similar condition in grabMouse as symbian after this commit. Task-number: QTBUG-5658 Reviewed-by: Jason Barron
* | Add edit menu support for phones with a 'pencil' key.Jason Barron2009-11-111-0/+1
|/ | | | | | | | | | | | | | | | | The native FEP handles the 'pencil' key on some platforms by opening an edit menu on the screen that contains things like 'Insert Symbol' and 'Writing Language' for text input widgets. This was previously not working in the Qt input method implementation because in order for the FEP system to open the menu, it must be able to access the menu bar (CBA). This is done my using the object provider mechanism (MOP) and an implementation of the MopNext() function was missing. Adding this and also setting the AppUi as the MOP parent for top level widgets allowed the FEP framework to find it's way to the menubar and thus open a menu. Task-number: QTBUG-5606 Reviewed-by: axis <qt-info@nokia.com> Reviewed-by: mread <qt-info@nokia.com>
* Merge branch '4.6' into mmfphononGareth Stockwell2009-11-031-68/+4
|\
| * Switched setWindowIcon_sys to use QPixmpa::toSymbianCFbsBitmap in S60.Janne Anttila2009-10-271-68/+4
| | | | | | | | | | | | | | | | | | | | | | There were TODOs in code to remove the temporary solution for creating native CFbsBitmap out of QPixmap. Now when QPixmpa has native backed and it provides toSymbianCFbsBitmap, it it preferred to use toSymbianCFbsBitmap since in best case it can only duplicate the bitmap handle instead of copying the data. Task-number: QTBUG-4948 Reviewed-by: Jani Hautakangas
* | When creating a Symbian WId for a visible widget, make the controlGareth Stockwell2009-10-231-1/+3
|/ | | | | | | | | | | | visible after activating the window. This change was required in order to be able to run the test case for the below task; however, it is more generally required. Without it, the contents of the descendents of this widget will not be visible, until they are explicitly hidden and then re-shown. Task-number: QTBUG-4787 Reviewed-by: axis
* Lowering toplevel widget puts app to background.Miikka Heikkinen2009-10-221-2/+7
| | | | | | | Since raising toplevel widget nowdays brings the whole app to top, logically lowering toplevel widget should put the app to background. Reviewed-by: axis
* Fixed QWidget::raise in SymbianMiikka Heikkinen2009-10-191-1/+6
| | | | | | | | If toplevel window is raised, the whole application is now raised to foreground. Task-number: QT-2162 Reviewed-by: axis
* Merge commit 'origin/4.6' into mmfphononFrans Englich2009-10-151-1/+6
|\ | | | | | | | | | | Conflicts: src/corelib/kernel/qcoreevent.cpp src/corelib/kernel/qcoreevent.h
| * Remove incorrect optimisation in S60 widget / control stack syncShane Kearns2009-10-081-3/+8
| | | | | | | | | | | | | | | | | | | | | | Incorrect optimisation - for popup windows, Qt's focus is moved before hide_sys is called, resulting in the popup window keeping its elevated position in the CONE control stack. This can result in keyboard focus being in an invisible widget in some conditions - e.g. QTBUG-4733 Task-number: QTBUG-4733 Reviewed-by: axis
* | When reparenting native widget, delayed deletion of Symbian controlGareth Stockwell2009-10-151-1/+12
| | | | | | | | | | | | | | | | | | | | | | until control returns to the event loop. This is necessary because reparenting can be triggered from the context of a control's event handler. If reparenting causes that control to be deleted, a crash can result. Task-number: QTBUG-4664 Reviewed-by: axis
* | Modified reparenting to correctly deal with native child widgetsGareth Stockwell2009-10-091-22/+47
|/ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Both reparenting and modification of window flags are done by calling QWidget::setParent. If either the parent changes, or a non-top-level widget becomes top-level (as a result of OR-ing Qt::Window into its window flags), a new native window ID is created for the widget. The Symbian implementation of setParent_sys had a flaw which manifested itself in the following situation: 1. We start with a native widget X, and its child Y, which is also native. There exist parent-child relationships between the associated CCoeControl instances, and the native windows (represented by RWindow handles). 2. X gets a new native window created as a result of a call to setParent. 3. QWidgetPrivate::reparentChildren calls SetParent on Y's control, to re-establish the parent-child relationship. The problem is that the window owned by Y's control now has no parent, so if we try to re-size the widget, the window server panics the client thread (WSERV-52). Because Symbian does not allow existing windows to be re-parented, and nor does it allow a window-owning control to re-create a new window, the only way to provide Y's window with a parent is to destroy the control and create a new one, passing in X's new window to the CCoeControl::CreateWindowL function. The changes made are as follows: a) QWidgetPrivate::reparentChildren is therefore modified to call create_sys, with destroyOldWindow set to true. b) QWidgetPrivate::create_sys is modified to take account of the value of this flag in all cases. (Previously it only did so if a new WId was passed in by the caller). c) The call to setWinId is delayed until the control and window are fully initialized. This is to allow us to emit a new event, WinIdChanged, from setWinId, in order to inform the widget that its winId has changed. d) QWidgetPrivate::activateSymbianWindow is modified in order to support this change of call ordering. Note that QWidgetPrivate::create_sys requires some re-factoring in order to remove the redundancy between the top-level and child widget cases. Task-number: QTBUG-4664 Reviewed-by: axis
* Merge commit 'qt/4.6' into mmfphononFrans Englich2009-10-021-47/+24
|\ | | | | | | | | Conflicts: src/gui/kernel/qwidget_s60.cpp
| * Revert "Fix for Symbian window activation/focus problem."Jani Hautakangas2009-10-011-59/+0
| | | | | | | | This reverts commit fafd16474aee5bbf47ee37e1ba739f3b3ceb9c33.
| * Fix for Symbian window activation/focus problem.Jani Hautakangas2009-10-011-0/+59
| | | | | | | | | | | | | | | | When dialog is shown on top of a top level window, the focus does not get back to top level window when dialog is closed. This is a fix for that. Reviewed-by: Shane Kearns
| * Merge branch 'focusFixes' into 4.6axis2009-09-301-45/+22
| |\ | | | | | | | | | | | | | | | Conflicts: src/gui/kernel/qwidget_p.h src/gui/kernel/qwidget_s60.cpp
| | * Fixed some focus issues on Symbian.axis2009-09-301-15/+21
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | There are really two bugs that are fixed in this commit: - SetFocus() in Symbian does not automatically clear focus on the previously focused control, so we have to remember that control and clear it ourselves. - Symbian assumes that it is always the control at the top of the control stack that should have focus, and if this isn't the case, focus may or may not work depending on whether Symbian has had a chance to reset the focus or not. Therefore, whenever we change focus on a control, we have to also readd that control to the top of the stack, to ensure that it stays focused. RevBy: Janne Anttila
| | * Revert "Switched to asynchronous focus handling on Symbian."axis2009-09-301-28/+0
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This reverts commit b6377f43410b14125a66ffd02acde69cfb6e455e. The asynchronous handling caused too many headaches with input methods, which expect the focus status to be updated immediately. This may break the test case that was originally fixed by this patch (I cannot find out which one at the moment), but that will have to be solved in a different way. Conflicts: src/corelib/kernel/qcoreevent.cpp src/corelib/kernel/qcoreevent.h src/gui/kernel/qwidget.cpp src/gui/kernel/qwidget_p.h src/gui/kernel/qwidget_s60.cpp
* | | Fixed indenting to comply with Qt code styleGareth Stockwell2009-10-021-15/+15
| | |
* | | Enable pointer events on native child widgetsGareth Stockwell2009-10-011-0/+5
| | |
* | | Immediately activate native windows created by calling winId() on an ↵Gareth Stockwell2009-10-011-4/+19
|/ / | | | | | | | | | | already-visible widget If a widget is visible when winId() is called on it, this change means that the newly-created native window will be activated immediately.
* | Removed trace statements from QtGuiGareth Stockwell2009-09-281-189/+8
| |
* | Removed calls to CCoeControl::DrawNow from QWidgetPrivate::setWSGeometryGareth Stockwell2009-09-281-30/+7
| |
* | Implemented QWidget::mapFromGlobal, QWidget::mapToGlobal, for case where ↵Gareth Stockwell2009-09-281-5/+8
| | | | | | | | child widget owns a native window
* | Revert "Revert all GUI changes from the MMF work."Frans Englich2009-09-281-51/+375
|/ | | | This reverts commit 9345d47c3945b61a27724508e8b3d0aaf7b57bcf.
* Compile fix for platforms prior to Symbian^3.Jason Barron2009-09-231-0/+2
| | | | | | | | | The advanced pointer events are only available on Symbian^3 and higher so we need to make sure these are protected by an #ifdef. We might have to re-factor this later into a plugin in order to get this running on older versions. Reviewed-by: axis
* Implement Qt::WA_AcceptTouchEvents on S60Bradley T. Hughes2009-09-231-0/+12
| | | | | | | | Turning this attribute ends up calling RWindow::EnabledAdvancedPointers(), which tells Symbian to send us multiple pointer events with extended info. Reviewed-by: Jason Barron
* Refactor soft keys implementation.Jason Barron2009-09-211-72/+0
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Reviewed-by: Alessandro Portale Squashed commit of the following: commit dae5eda6996d48c12c4a5efd3f6042eb293bacbf Author: Jason Barron <jbarron@trolltech.com> Date: Fri Sep 18 10:32:26 2009 +0200 Only update soft keys when KEYPAD_NAVIGATION is enabled. For 4.6, let's just call these functions when keypad navigation is defined to minimize the impact on other platforms. They should probably get thier own define some day. commit 30a730553531f0f138de5eddb30413936a34fa36 Author: Jason Barron <jbarron@trolltech.com> Date: Fri Sep 18 10:30:23 2009 +0200 Add/remove the menu bar action when menu bar is added/removed. commit a83343a2870b34c228c8bc5e6274607b0e97baf6 Author: Jason Barron <jbarron@trolltech.com> Date: Fri Sep 18 10:28:55 2009 +0200 Compile fix for other platforms commit 39c9e3a0a1d3d62bf6ebd3212cfd2a81b86b9b2a Author: Jason Barron <jbarron@trolltech.com> Date: Thu Sep 17 21:37:59 2009 +0200 Fix 'softkeys' example after API re-factoring. Clean up this example and use the simplified soft key API. Now the actions are only allocated in the constructor and dynamically updated by calling addAction and removeAction. commit 314e45c33f40552db74e61755c2a3b0f8c77a41a Author: Jason Barron <jbarron@trolltech.com> Date: Thu Sep 17 21:30:32 2009 +0200 Re-factor and simplify the soft keys API. Several things here: - Move all the logic into QSoftKeyManager - Move the files into 'kernel' since it is not a widget - Remove QWidget::setSoftKey*(). Use addAction/removeAction instead - Made soft keys update on focus, window activation, and action changes. - Fixed several memory leaks where QAction's were created too often - QAction ownership pushed out to widget's - Added Select/Cancel soft keys for comboboxes and menus to be more consistent to platform look-and-feel. commit fb4c240d970262e9872dc5737df6808879143c75 Author: Jason Barron <jbarron@trolltech.com> Date: Mon Sep 7 15:49:31 2009 +0200 Merge the Symbian implementation with the other platforms nativeMenuBar It seems this has been refactored to share more code across the various platforms that support native menubars so the Symbian code can be mostly removed. commit aa55e4bcd1f009ab35c9519e18aa325fd212dd23 Author: Jason Barron <jbarron@trolltech.com> Date: Wed Aug 26 17:00:34 2009 +0200 Change filenames and move softkey stuff from 'widgets' to 'kernel'. This thing isn't a widget and therefore should not be in the 'widgets' subdirectory of gui. Also rename the files in preparation for refactoring and extending.
* Enable compiling Symbian port with QT_NO_CURSORShane Kearns2009-09-171-1/+2
| | | | | | | Added some missing #ifdef QT_NO_CURSOR, so the symbian port still compiles if this feature is configured out. Reviewed-by: Jason Barron
* QCursor support for Symbian OSShane Kearns2009-09-151-19/+58
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Reviewed-By: Jason Barron Reviewed-By: Alessandro Portale Summary: QT_NO_CURSOR is now not defined for symbian builds Existing QCursor APIs are all supported New public API, QApplication::setNavigationMode, to allow the navigation mode to be set. I.E. on an S60 3.2 phone, some applications will want a virtual mouse cursor (web browser), while others are designed for keypad navigation. Symbian HAL is used for detecting input capabilities. Fix DND, code cleanup & comment QCursor visibility now uses a refcount, and is called from DND and the setNavigationMode so they are both simpler and don't interfere with each other. QApplication::setNavigationMode New public API for configuring cursor/keypad navi style. This links in with ongoing work on the 4-way keypad navi branch, but 2-way and 4-way modes both act as 2-way mode until that is integrated Some of the demos/examples have cursor switched on (those that were not usable with keypad) Virtual mouse support for non touch, non mouse phones (tested on N78) add *.d and .metadata (carbide debug file / workspace dir) to .gitignore System pointers are unavailable when using sprite workaround, so the system cursor shapes are compiled into qtgui as resources. MAC port does this also for shapes that aren't standard on the MAC. Refactor Drag'n'Drop to use QCursor Add test case to check all system cursor shapes Simply a mainwindow containing a label widget for each cursor shape, with the cursor property set appropriately QCursor(QBitmap,QBitmap) supported Fixed problem with the image & mask being inverted when using the QCursor constructor that takes two mono bitmaps. add .make.cache files to .gitignore Correct implementation of QApplication::setOverrideCursor QApplication::restoreOverrideCursor and QApplication::setOverrideCursor are now working correctly on Symbian platform. Performance will be slower compared with other platforms, because the Symbian window server has a cursor associated with each native window. Add test case for custom cursors Create a pixmap cursor and associate it with a widget. No changes to production code, since test passed 1st time ;) Add manual test for QCursor Make cursor independent of construction order Updated to work around window server issue where contruction order affects what cursor is displayed in child windows. Also changed to effectiveWinId following review comments Also fixed a problem which would make qcursor not link if configured with QT_NO_CURSOR Moved some multiply declared extern functions from cpp to _p.h files Implemented Symbian versions of the cursor functions. Merged in work I'd done based on tower. Fill in bits of stub functions based on windows port Removed QT_NO_CURSOR from list of config options forced on symbian Recompiled configure.exe Added stub functions for the missing functions in s60 port
* Fixed QWidget::activateWindow for Symbian.Janne Anttila2009-09-111-6/+5
| | | | | | | | | | | | | | activeWindow documentation says: "active window is a visible top-level window that has the keyboard input focus" and "If you want to ensure that the window is stacked on top as well you should also call raise(). Note that the window must be visible, otherwise activateWindow has no effect." What we were doing earlier was basically raise. Now we just set the keyboard focus to underlying native window. Task-number: 260685 Reviewed-by: Jason Barron
* Fixed QWidget::setWindowState for Symbian.Janne Anttila2009-09-111-16/+27
| | | | | | | | | | | | | | | | | | | | | | | | | | Before this commit calling setWindowState(Qt::WindowFullScreen) on a widget instance affected all new widget instances created after this method. This bug happened due to fact that window decorations i.e. statuspane and softkeys visibility was only changed when switching to or from fullscreen state. In the reported bug it happened that second widget was initially in Qt::WindowNoState and it was changed to Maximized. Since window decorations are global not window specific at the moment, the default decoration visibility for second window is the one to which previous window has set them. In this case previous window was in fullscreen and that's why the decorations were visible also for second maximized window. Probably the right fix would be to change the decoration to window specific but that is quite a big change and for now the bug is fixed with this commit. Autotest: Excluding new test case, same results before and after. Task-number: 261048 Reviewed-by: Jason Barron
* Update license headers again.Jason McDonald2009-09-091-4/+4
| | | | Reviewed-by: Trust Me
* Merge branch 'minimizeWrapperApp' into 4.6axis2009-09-031-1/+3
|\
| * Move the S60/Avkon framework initialization into QtGui.axis2009-09-031-1/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This way we avoid having a lot of code in a static (and unmaintainable) library. The s60main static library now currently has only one task: to call main(). To move the initialization into QtGui also meant a change in how the S60 framework is created, because we can no longer use the trick where we create and start the the S60 event loop and then have the framework call us back to start main(). The initialization now follows the creation and destruction of QApplication, which is a lot more in line with how other platforms do it. Since S60 doesn't support creating the environment, and *then* starting it (both are executed by the same call), we had to open up the S60 framework construction classes and just mirror what they do. This means that after QApplication construction is done, the S60 framework is initialized, but nothing will run yet and control will return to main(), where the user can start the event loop himself. One of the quirks of this approach is that the construction of the S60 framework makes a new cleanup stack. This means that any active traps will not be active anymore, and leaving without setting a new trap will most likely panic. This shouldn't be a problem for us, since Qt is never supposed to leave, but it means that if anyone uses the cleanup stack without setting a new trap, they will receive a panic. It was considered to add a trap mark in QApplication construction and then removing it on destruction, but it was dropped because leaving from main() is still undefined (even if the old cleanup stack would be restored in the destructor, we wouldn't be able to stop the exception from unwinding the stack, and the cleanup stack would then be unbalanced). RevBy: Jason Barron RevBy: Janne Anttila AutoTest: QWidget passed with same failure count
* | Update tech preview license header for files that are new in 4.6.Jason McDonald2009-08-311-13/+13
|/ | | | Reviewed-by: Trust Me
generated by cgit v0.12 at 2024-12-29 03:50:58 (GMT)