summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Merge branch '4.7' of scm.dev.nokia.troll.no:qt/qt-s60-public into ↵Qt Continuous Integration System2011-04-1320-109/+262
|\ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 4.7-integration * '4.7' of scm.dev.nokia.troll.no:qt/qt-s60-public: (21 commits) Have the backing store destroyed also in special Symbian scenarios. Use FixNativeOrientation on Symbian when the application wants it. QS60Style: itemview item with checkbox and text is drawn incorrectly Revert "Have the backing store destroyed also in special Symbian scenarios." Add focus frame support in style sheet Two QListView autotests do not pass on Symbian^3 releases tst_QStyle::drawItemPixmap test case fails on Symbian^3 Add the auto detection for OpenVG in configure Add focus frame support in stylesheet Make QS60Style to support new Tab graphics in new Symbian releases GraphicsView is not reset if focusItem is not set when keyboard closes Check the validity of qt_desktopWidget before dereferencing Labels are not visible in dialogs with all themes (part2) Labels are not visible in dialogs with all themes Enable webkit for symbian-gcce in configure Enable webkit build for the makefile build system with gcce Add the rules for gcce in do_not_build_as_thumb.prf Correct some parameters for the makefile build sytem, armcc and gcce Support QMAKE_LFLAGS.ARMCC and QMAKE_LFLAGS.GCCE in makefile build system Do not unnecessarily draw application twice in landscape ...
| * Have the backing store destroyed also in special Symbian scenarios.Laszlo Agocs2011-04-121-0/+6
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The backing store tracker's registerWidget and unregisterWidget functions are called when EPartiallyVisible and ENotVisible events come from WSERV. However if an application sends its window group to background right after caling show() and before entering the event loop, there is a chance that all the application will receive is an ENotVisible event, leading to calling unregisterWidget() without a previous registerWidget(). In this case the backing store was not destroyed and the application was consuming GPU memory even while it was staying in background. This patch ensures registerWidget() is called always before unregisterWidget() in the Symbian-specific event handler for ENotVisible, and replaces the previous patch, a12d41076919a133e63de63dff5c1a131a0564e4, which caused regression in autotests. Task-number: QTBUG-18493 Reviewed-by: Gareth Stockwell
| * Use FixNativeOrientation on Symbian when the application wants it.Laszlo Agocs2011-04-125-47/+132
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Applications can request the usage of RWindow::FixNativeOrientation() by setting the new Qt::WA_SymbianNoSystemRotation attribute for their fullscreen top-level widget (which in practice will either be a graphics or declarative view). This will fix the underlying EGL window surface and the QWidget dimensions to the dimension of the native orientation of the device (typically portrait). The default auto-rotation can be left enabled, however it will be up to the application to rotate the drawing. Global notifications, VKB, etc. will still appear in the proper orientation. Another benefit is improved performance in the non-native orientation. Task-number: QTBUG-17742 Reviewed-by: Jason Barron Reviewed-by: Jani Hautakangas
| * QS60Style: itemview item with checkbox and text is drawn incorrectlySami Merila2011-04-121-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | QS60Style deduces incorrectly when itemview only contains checkbox. This is minor, yet highly annoying bug that prevents highlighted itemview item from showing its content with themes that have opaque itemview item highlight (highlight covers the text that is underneath). As a fix, check itemview item text and icon content before declaring it as "checkbox only". Task-number: QTBUG-18694 Reviewed-by: Tomi Vihria
| * Revert "Have the backing store destroyed also in special Symbian scenarios."Laszlo Agocs2011-04-121-2/+1
| | | | | | | | This reverts commit a12d41076919a133e63de63dff5c1a131a0564e4.
| * Merge branch '4.7' of scm.dev.nokia.troll.no:qt/qt-s60-public into 4.7Sami Merila2011-04-121-2/+4
| |\
| | * Add focus frame support in style sheetGuoqing Zhang2011-04-121-2/+4
| | | | | | | | | | | | | | | Task-number: QTBUG-16207 Reviewed-by: Sami Merila
| * | Two QListView autotests do not pass on Symbian^3 releasesSami Merila2011-04-122-3/+2
| |/ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | tst_QListView::taskQTBUG_2678_spacingAndWrappedText() fails because: QS60Style adds to the itemview item content size margins and empty space, which shouldn't be part of the content size. As a fix, remove these. taskQTBUG_435_deselectOnViewportClick() fails because: Sending a click to a selected itemview item, when it should be sent to below that specific one. It was using hardcoded (center + 20) and autotest assumed that this would be outside of first item. In S60 with touch support, the itemview items are rather tall (49 pixels). As a fix, autotest now uses calculated value, which ensures that click is sent to outside first item. Task-number: QT-4810 Reviewed-by: Tomi Vihria
| * tst_QStyle::drawItemPixmap test case fails on Symbian^3Sami Merila2011-04-111-0/+12
| | | | | | | | | | | | | | | | | | | | | | The autotest assumes that created QPixmap are exactly same irregardless how the pixmaps were created. However, there is no guarantee that pixmaps that look the same (i.e. in this case green rectangles) are "same". QPixmap is platform dependent and might have e.g. optimized format in some of the platforms. Task-number: QT-4805 Reviewed-by: Jani Hautakangas
| * Add the auto detection for OpenVG in configureLiang Qi2011-04-081-3/+3
| | | | | | | | | | Task-number: QTBUG-18647 Reviewed-by: axis
| * Add focus frame support in stylesheetGuoqing Zhang2011-04-081-0/+5
| | | | | | | | | | Task-number: QTBUG-16027 Reviewed-by: Sami Merila
| * Make QS60Style to support new Tab graphics in new Symbian releasesSami Merila2011-04-082-21/+42
| | | | | | | | | | | | | | | | | | | | | | In SR.11 there is a new theme graphics (with separate ID) for TabWidget tab shapes. Also, shapes are no longer set overlapped, but are side-by-side without any gaps. Earlier releases still use the existing graphics with overlapping. Task-number: QT-4762 Reviewed-by: Miikka Heikkinen
| * Merge branch '4.7' of scm.dev.nokia.troll.no:qt/qt-s60-public into 4.7Sami Merila2011-04-081-2/+4
| |\
| | * Check the validity of qt_desktopWidget before dereferencingRuth Sadler2011-04-071-2/+4
| | | | | | | | | | | | | | | Task-number: QTMOBILITY-1494 Reviewed-by: Gareth Stockwell
| * | GraphicsView is not reset if focusItem is not set when keyboard closesSami Merila2011-04-081-6/+8
| |/ | | | | | | | | | | | | | | | | | | | | | | When splitview closes (i.e. keyboard is dismissed), application's graphicsview containing the input widget (focus item) is reset back to original graphicsview transformation. As a fix, resetTransform is always called, when rootItem is found from graphicsView. Task-number: QT-4858 Reviewed-by: Guoqing Zhang
| * Labels are not visible in dialogs with all themes (part2)Sami Merila2011-04-071-0/+1
| | | | | | | | | | | | | | Added include to the previous fix to prevent compilation issue. Task-number: QT-4559 Reviewed-by: Tomi Vihria
| * Labels are not visible in dialogs with all themesSami Merila2011-04-071-0/+7
| | | | | | | | | | | | | | | | | | | | | | | | | | QS60Style uses different theme background for dialogs. However, labels use just one color irregardless of background texture. Therefore, in certain themes, this might produce an issue, where foreground color (label) is not visible from background. To fix this, set the label color to correct theme color when label is contained within a dialog. Task-number: QT-4559 Reviewed-by: Tomi Vihria
| * Enable webkit for symbian-gcce in configureLiang Qi2011-04-071-1/+0
| | | | | | | | Reviewed-by: axis
| * Enable webkit build for the makefile build system with gcceLiang Qi2011-04-072-10/+9
| | | | | | | | | | | | | | Task-number: QTBUG-18484 Task-number: WEBKIT-57841 Reviewed-by: axis Reviewed-by: Janne Koskinen
| * Add the rules for gcce in do_not_build_as_thumb.prfLiang Qi2011-04-071-0/+5
| | | | | | | | Reviewed-by: axis
| * Correct some parameters for the makefile build sytem, armcc and gcceLiang Qi2011-04-073-12/+9
| | | | | | | | Reviewed-by: axis
| * Support QMAKE_LFLAGS.ARMCC and QMAKE_LFLAGS.GCCE in makefile build systemLiang Qi2011-04-072-1/+4
| | | | | | | | Reviewed-by: axis
| * Do not unnecessarily draw application twice in landscapeMiikka Heikkinen2011-04-072-1/+9
| | | | | | | | | | | | | | | | | | | | | | | | Fullscreen with softkeys dialogs would draw twice when orientation was changed to landscape, causing flicker. This happened because status pane change event and layout change event both caused redraw with different client area size. Added a check to avoid handling client area change on account of status pane change, if status pane is not visible and there was no change in its visibility. Task-number: QTBUG-18496 Reviewed-by: Sami Merila
| * Have the backing store destroyed also in special Symbian scenarios.Laszlo Agocs2011-04-071-1/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The backing store tracker's registerWidget and unregisterWidget functions are called when EPartiallyVisible and ENotVisible events come from WSERV. However if an application sends its window group to background right after caling show() and before entering the event loop, there is a chance that all the application will receive is an ENotVisible event, leading to calling unregisterWidget() without a previous registerWidget(). In this case the backing store was not destroyed and the application was consuming GPU memory even while it was staying in background. This patch makes unregisterWidget() not to check the widget's presence in the m_widgets set, instead the condition for deleting the backing store is solely an empty set. Task-number: QTBUG-18493 Reviewed-by: Gareth Stockwell
* | Merge branch '4.7' of scm.dev.nokia.troll.no:qt/qt-qml into 4.7-integrationQt Continuous Integration System2011-04-127-2/+119
|\ \ | | | | | | | | | | | | | | | | | | * '4.7' of scm.dev.nokia.troll.no:qt/qt-qml: ListView has wrong keyPressEvent behaviour when vertical Ensure view is positioned correctly when orientation changes. Fix regression in wigglytext.qml
| * | ListView has wrong keyPressEvent behaviour when verticalMartin Jones2011-04-121-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | Regression intorduced by RTL changes. Change-Id: I272d07cd21d04f3e534aa183b1b10fcc8d062b79 Task-number: QTBUG-18581 Reviewed-by: Bea Lam
| * | Ensure view is positioned correctly when orientation changes.Martin Jones2011-04-123-0/+56
| | | | | | | | | | | | | | | | | | Change-Id: I7fbedff965ae8c89dcbb96ba5dcee85c07aa29b1 Task-number: QTBUG-17065 Reviewed-by: Bea Lam
| * | Fix regression in wigglytext.qmlMichael Brasser2011-04-124-0/+61
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This change re-adds the code removed in 8e9c28eaa4d7a3372b9a9a21a984701b62f96456 (which caused this regression), while keeping the new code as well (to specially handle the case of registration in componentCompleted()). Change-Id: I707e3d2ead9ea25079f79cd5e5886d1dc1c69d1b Task-number: QTBUG-18362 Reviewed-by: Aaron Kennedy
* | | Merge branch '4.7' of scm.dev.nokia.troll.no:qt/oslo-staging-1 into ↵Qt Continuous Integration System2011-04-1223-86/+117
|\ \ \ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 4.7-integration * '4.7' of scm.dev.nokia.troll.no:qt/oslo-staging-1: look for makespec in the build dir first make collection of SUBDIRS in solution generator less bizarre run depend_command in build dir in VS generators as well fixify target source against build tree fix fixifying of QMAKE_SUBSTITUTES dist target: fixify OBJECTS_DIR against output dir fix paths of vpath-resolved files simplify: absolute fixification ignores the base dir arguments simplify: fileFixify for all same paths is always "." simplify: the input and output dirs are already normalized stop fixifying after first success create a pwd string with a trailing slash only on demand make uic3-generated moc files depend on mocinclude.tmp useful location reporting for errors from QMAKE_SUBSTITUTES clean up visual studio identification code prefix TEMPLATE_PREFIX to TEMPLATE even if it is "default-constructed" fix detection of relative location information
| * | | look for makespec in the build dir firstOswald Buddenhagen2011-04-121-3/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | it's more natural to look into the build dir before the source dir, and it's what the qmake-generated makefiles mean when re-invoking qmake. specifically, this works around the problem that relative paths with excess ".."s pointing below the root are happily ignored and thus truly bizarre makespec paths may be constructed by the qmake re-invocations if the source dir is less nested than the build dir. Task-number: QTBUG-9817 Reviewed-by: mariusSO
| * | | make collection of SUBDIRS in solution generator less bizarreOswald Buddenhagen2011-04-122-21/+22
| | | | | | | | | | | | | | | | | | | | | | | | | | | | de-duplicate code, and on the way don't try to re-resolve project variables of subprojects against the contents of the top level project. Reviewed-by: mariusSO
| * | | run depend_command in build dir in VS generators as wellOswald Buddenhagen2011-04-124-4/+18
| | | | | | | | | | | | | | | | | | | | | | | | followup to b139e7e96e5c47b412c4f0bbc4ae11d5cca99e61 Reviewed-by: mariusSO
| * | | fixify target source against build treeOswald Buddenhagen2011-04-121-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | that's where one would expect a target, after all. affects only extra targets explicitly requesting fixification, i.e., nothing. Reviewed-by: mariusSO
| * | | fix fixifying of QMAKE_SUBSTITUTESOswald Buddenhagen2011-04-121-6/+10
| | | | | | | | | | | | | | | | | | | | | | | | | | | | use the correct bases. notably, don't expect the input file in the output dir. Reviewed-by: mariusSO
| * | | dist target: fixify OBJECTS_DIR against output dirOswald Buddenhagen2011-04-121-1/+2
| | | | | | | | | | | | | | | | | | | | | | | | the thing is terminally broken anyway, but whatever Reviewed-by: mariusSO
| * | | fix paths of vpath-resolved filesOswald Buddenhagen2011-04-121-4/+3
| | | | | | | | | | | | | | | | | | | | Task-number: QTBUG-8169 Reviewed-by: mariusSO
| * | | simplify: absolute fixification ignores the base dir argumentsOswald Buddenhagen2011-04-121-1/+1
| | | | | | | | | | | | | | | | Reviewed-by: mariusSO
| * | | simplify: fileFixify for all same paths is always "."Oswald Buddenhagen2011-04-121-3/+1
| | | | | | | | | | | | | | | | | | | | | | | | this must have been the most arcane way to generate a single dot ever Reviewed-by: mariusSO
| * | | simplify: the input and output dirs are already normalizedOswald Buddenhagen2011-04-121-1/+1
| | | | | | | | | | | | | | | | Reviewed-by: mariusSO
| * | | stop fixifying after first successOswald Buddenhagen2011-04-121-0/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | somewhat unlikely that this had much real-world effects ... except eating yet more cpu. Reviewed-by: mariusSO
| * | | create a pwd string with a trailing slash only on demandOswald Buddenhagen2011-04-121-6/+7
| | | | | | | | | | | | | | | | Reviewed-by: mariusSO
| * | | make uic3-generated moc files depend on mocinclude.tmpOswald Buddenhagen2011-04-121-0/+1
| | | | | | | | | | | | | | | | | | | | Task-number: QTBUG-16951 Reviewed-by: mariusSO
| * | | useful location reporting for errors from QMAKE_SUBSTITUTESOswald Buddenhagen2011-04-123-1/+16
| | | | | | | | | | | | | | | | Reviewed-by: mariusSO
| * | | clean up visual studio identification codeOswald Buddenhagen2011-04-1212-22/+16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | TEMPLATE is (now) reliably prefixed with "vc" when a vs generator is used. Reviewed-by: mariusSO
| * | | prefix TEMPLATE_PREFIX to TEMPLATE even if it is "default-constructed"Oswald Buddenhagen2011-04-121-7/+7
| | | | | | | | | | | | | | | | | | | | | | | | | | | | that way prf files don't have to check both the prefix and the actual template to identify visual studio mode. Reviewed-by: mariusSO
| * | | fix detection of relative location informationOswald Buddenhagen2011-04-111-5/+7
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | an empty line number does not indicate relative loc info - it may be the result of -no-ui-lines. instead, an empty file name does indicate it - no file name at all makes no sense, so this means a previous messages has set it already. and we need this additional detection, as the entire ts file may have no line number info to base the decision on at all.
* | | | Merge branch '4.7' of scm.dev.nokia.troll.no:qt/qt-qml into 4.7-integrationQt Continuous Integration System2011-04-1210-31/+136
|\ \ \ \ | | |/ / | |/| | | | | | | | | | | | | | * '4.7' of scm.dev.nokia.troll.no:qt/qt-qml: Allow enum values to be used as signal parameters ListModel::clear() should not clear roles
| * | | Allow enum values to be used as signal parametersBea Lam2011-04-128-28/+94
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The fix does not work for enums declared outside of the class that emits the signal, since in this case it is not possible to access the metaobject of the class that declared the enum. However the fix makes a special case for the Qt namespace to allow enums in this namespace to be used. Task-number: QTBUG-15983 Change-Id: I6f72255b07010311a20fe068bd97c7db7b294d9b Reviewed-by: Aaron Kennedy
| * | | ListModel::clear() should not clear rolesBea Lam2011-04-122-3/+42
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Clearing a model and then appending a new object with a subset of the previous roles was causing a crash since ListModel cleared the roles and VisualDataModel did not, so VisualDataModel was requesting invalid roles. This could have been fixed by clearing the meta object and its property cache in in VisualDataModel but this is less efficient, and also the general use case is for model data to always have the same roles. Task-number: QTBUG-18587 Change-Id: Ib11d2292888ab7a41e772b1e11700cd665e94ae7 Reviewed-by: Michael Brasser
* | | | Merge branch '4.7' of scm.dev.nokia.troll.no:qt/qt-water-staging into ↵Qt Continuous Integration System2011-04-111-1/+1
|\ \ \ \ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 4.7-integration * '4.7' of scm.dev.nokia.troll.no:qt/qt-water-staging: Cocoa: Fixup wrong code in patch baaa5ae
generated by cgit v0.12 at 2025-01-17 19:48:36 (GMT)