summaryrefslogtreecommitdiffstats
path: root/src
Commit message (Collapse)AuthorAgeFilesLines
* Whitespace cleanup.Jan-Arve Sæther2009-10-191-13/+13
|
* API addition: Make it possible to set the size policy of an anchor.Jan-Arve Sæther2009-10-164-54/+88
| | | | | | | | | | | | | | | | | | | | | | | | | | After this commit, when you modify the spacing of an anchor you are effectively modifying the preferred size of the anchor, since all anchors (except internal ones) have their minimumSizeHint to 0 and maximumSizeHint to QWIDGETSIZE_MAX. I also changed the sizeHintsFromItem to be more generic so that I could use it for anchors. (Thus, it was renamed to "internalSizeHints"). It now only takes care of setting the min/pref/exp/maxSize of AnchorData based on the anchor/item size hint and their size policies. As a consequence of all of this, setFixedSize changed behaviour and became setPreferredSize (since setSpacing is basically setPreferredSize). The implementation of that now only sets the prefSize. The patch also has an unrelated fix for IgnoreFlag, where it will (again) return the minimumSize for sizeHint(Qt::PreferredSize) instead of the maximumSize. This was to be more consistent with qgridlayoutengine.cpp. The docs are not very clear on this behaviour unfortunately. This API change has been discussed. Reviewed-by: alexis
* Merge branch 'fixes' of ↵Jan-Arve Sæther2009-10-161-159/+165
|\ | | | | | | | | | | | | | | | | git://gitorious.org/~fleury/qt/fleury-openbossa-clone into 4.6 Conflicts: src/gui/graphicsview/qgraphicsanchorlayout_p.cpp src/gui/graphicsview/qgraphicsanchorlayout_p.h tests/auto/qgraphicsanchorlayout/tst_qgraphicsanchorlayout.cpp
| * QGAL: refactor and document the simplification algorithmCaio Marcelo de Oliveira Filho2009-10-151-159/+165
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Refactor the simplifyGraphIteration() function. The aim was to make it more clear without changing too much the way it does stuff. Before it collected a list of candidates and then filtered that into sublists of same order, and after that removed the center edges from the borders. It also reversed the list if the direction wasn't forward, to pass it in forward order to simplifySequentialChunk() helper function. The refactored version - take in account the order when building the candidates, this avoids index manipulation; - try to calculate 'beforeSequence' and 'afterSequence' as it builds the candidates list; - make simplifySequentialChunk() aware of directions, now it deals internally with reversed direction sequences. This commits also adds explanations to trickier parts of the code. Signed-off-by: Caio Marcelo de Oliveira Filho <caio.oliveira@openbossa.org> Reviewed-by: Artur Duque de Souza <artur.souza@openbossa.org> Reviewed-by: Eduardo M. Fleury <eduardo.fleury@openbossa.org>
| * QGAL: refactor calculateGraphs() methodCaio Marcelo de Oliveira Filho2009-10-152-75/+93
| | | | | | | | | | | | | | | | | | | | | | | | | | Create calculateTrunk() and calculateNonTrunk() methods, that calculate sizes for anchors in different parts of the simplified graph (using simplex when needed). Also fixes a minor leak when the nontrunk part is non-feasible. The old code left the loop leaving the contents of 'sizeHintConstraints' not allocated. Signed-off-by: Caio Marcelo de Oliveira Filho <caio.oliveira@openbossa.org> Reviewed-by: Artur Duque de Souza <artur.souza@openbossa.org>
| * QGAL: update the sizes of all anchors instead of doing it separatedlyCaio Marcelo de Oliveira Filho2009-10-152-9/+20
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | After calculations, update the size of all anchors in the simplified graph. Those updates were happening locally after each calculation (trunk and semifloats), however some anchors that were not involved in simplex calculation were missing. One concrete consequence of the previous behaviour is that semifloat parts that were simplified into just one anchor, didn't have the chance to set their sizeAt* values. One consequence of the new behaviour is one more test passing. Signed-off-by: Caio Marcelo de Oliveira Filho <caio.oliveira@openbossa.org> Reviewed-by: Artur Duque de Souza <artur.souza@openbossa.org>
| * QGAL: avoid having to re-calculate lists of variablesCaio Marcelo de Oliveira Filho2009-10-152-31/+27
| | | | | | | | | | | | | | | | | | | | | | | | | | Change the functions solvePreferred() and solveExpanding() to take a list of variables, so they don't need to calculate them based on the list of constraints. That way, the trunk variables are calculated only once. This commit also reduce the scope of 'sizeHintConstraints' variable instead of clearing and reusing it. Signed-off-by: Caio Marcelo de Oliveira Filho <caio.oliveira@openbossa.org> Reviewed-by: Artur Duque de Souza <artur.souza@openbossa.org>
| * QGAL: add a way to test whether simplex was used in a calculationCaio Marcelo de Oliveira Filho2009-10-152-0/+12
| | | | | | | | | | | | | | | | | | | | This is one good way to track whether simplification is doing all its job or not. However it can only track cases where the graph simplify to only one anchor. For more complex cases the graph dumps are the way to go. Signed-off-by: Caio Marcelo de Oliveira Filho <caio.oliveira@openbossa.org> Reviewed-by: Artur Duque de Souza <artur.souza@openbossa.org>
| * QGAL: improve dump graph and add helper codeCaio Marcelo de Oliveira Filho2009-10-152-3/+13
| | | | | | | | | | | | | | | | | | The graph dumper function take a name, and added debug code (that need to be enabled manually) to generate dumps from the graph before and after calculation. This is useful to debug simplification. Signed-off-by: Caio Marcelo de Oliveira Filho <caio.oliveira@openbossa.org> Reviewed-by: Artur Duque de Souza <artur.souza@openbossa.org>
| * QGraphicsAnchorLayout: Fix sequential anchor distribution w/ expanding sizesEduardo M. Fleury2009-10-151-86/+76
| | | | | | | | | | | | | | | | | | | | | | | | After the addition of expanding SizePolicy, the distribution of sequential anchors became wrong. We must now account for three intervals of distribution instead of only two. This commit also unifies the logic used by the sequential group anchor and the edge interpolator. Signed-off-by: Eduardo M. Fleury <eduardo.fleury@openbossa.org> Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@openbossa.org>
| * QGraphicsAnchorLayout: Ensure spacing is not negativeEduardo M. Fleury2009-10-152-0/+33
| | | | | | | | | | | | | | | | | | | | | | Currently we do not support negative-sized anchors in the graph, instead we invert such anchors and make their value positive. We consider changing this sometime in the future but until then, spacing must be non-negative at all times. Signed-off-by: Eduardo M. Fleury <eduardo.fleury@openbossa.org> Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@openbossa.org>
* | Merge branch '4.6' of scm.dev.nokia.troll.no:qt/qt-widget-team into 4.6Jan-Arve Sæther2009-10-162-2/+5
|\ \
| * | Fixed typo in qtableview.cpp.Gabriel de Dietrich2009-10-161-1/+1
| | | | | | | | | | | | | | | | | | Auto-test included. Will timeout on fail. Reviewed-by: Alexis
| * | Fixed a rare crash in qt_x11_enforce_cursorGabriel de Dietrich2009-10-151-1/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | A QPointer was set to point to a QWidget by one of its children, during its deletion. This happens during the child deletion, and after the call to QObject::clearGuards(), which means that the QPointer becomes a dangling one. The fix ensures that qt_x11_enforce_cursor will not be called with a being-deleted QWidget. The included auto-test doesn't test anything, except that it doesn't crash. Reviewed-by: Olivier
* | | Make sure that the anchor layout autotests pass in release configurationJan-Arve Sæther2009-10-162-2/+2
|/ / | | | | | | | | | | | | The problem was that lastCalculationUsedSimplex was only compiled in debug mode. The autotests run in release, so it did not compile. Reviewed-by: alexis
* | Merge branch '4.6' of scm.dev.nokia.troll.no:qt/qt-widget-team into 4.6Jan-Arve Sæther2009-10-1575-22092/+19853
|\ \
| * | Fixed updating problems QListView with IconModeGabriel de Dietrich2009-10-151-1/+1
| | | | | | | | | | | | | | | | | | | | | The updated rectangle coordinates didn't use the viewport offsets. Reviewed-by: Olivier Task-number: QTBUG-4819
| * | Merge commit origin/4.6 into team-widgets/4.6Olivier Goffart2009-10-1510-78/+113
| |\ \ | | |/
| | * Update IBY file to contain all Qt modules, inc. phonon, multimedia, sqlIain2009-10-141-0/+8
| | | | | | | | | | | | Reviewed-by: Janne Koskinen
| | * Cocoa: QInputDialog autotest reveals event dispatcher bugRichard Moe Gustavsen2009-10-141-8/+13
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | On Cocoa, we sometimes need to block sending posted events (because we need to flush the event que now and then without touching Qt events). But we forgot to do same for timer callback. So this patch makes sure that we dont send the timer event immidiatly if we are just flushing the event que. Rev-By: brad
| | * doc: Replaced \obsolete with \warning.Martin Smith2009-10-143-59/+76
| | | | | | | | | | | | | | | | | | Warning: This class is provided only to get access to S60 specific functionality in the application framework classes. It is not portable. We strongly recommend against using it in new applications.
| | * Merge branch '4.6' of git@scm.dev.troll.no:qt/qt into 4.6Miikka Heikkinen2009-10-141-0/+3
| | |\
| | | * Cocoa: qfiledialog will not exit cleanlyRichard Moe Gustavsen2009-10-141-0/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The native filedialog will not exit if it is told to hide. To remedy this, we just add an extra interrupt call so to inform the event dispatcher that it needs to return the the event loop to check if it has been told to quit Rev-By: olivier
| | * | Removed platform specific include from qprocess.hMiikka Heikkinen2009-10-144-11/+13
| | |/ | | | | | | | | | | | | | | | | | | | | | It wasn't necessary to have a Symbian specific include in qprocess.h, as 64-bit integer will work as pid also in Symbian. Task-number: QT-2266 Reviewed-by: Janne Anttila
| * | Merge origin/4.6 into team-widgets/4.6Olivier Goffart2009-10-1462-22012/+19733
| |\ \ | | |/
| | * Improve scissor clipping in the OpenVG engineRhys Weatherley2009-10-131-0/+4
| | | | | | | | | | | | | | | | | | | | | | | | If the scissor is being used to clip instead of the mask, use the bounding rectangle of the clip path if it cannot be handled as a simple rectangle. Reviewed-by: trustme
| | * Merge branch '4.6' of git@scm.dev.nokia.troll.no:qt/qt into 4.6Rhys Weatherley2009-10-131-1/+7
| | |\
| | | * Don't create too many window surfaces in DFBAnders Bakken2009-10-131-1/+5
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Since DirectFB handles the composition for us we don't need to invalidate anything when moving a window. Only on resize do we want another paint event. Reviewed-by: Donald Carr <donald.carr@nokia.com>
| | | * Make sure we don't repaint when moving a windowAnders Bakken2009-10-131-0/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | We need to make sure the QDirectFBWindowSurface returns true for isBuffered(). Otherwise QWSWindowSurface will force a repaint when moved. Reviewed-by: Donald Carr <donald.carr@nokia.com>
| | * | Explicitly request the alpha mask in the EGL configurationRhys Weatherley2009-10-131-0/+4
| | |/ | | | | | | | | | | | | | | | | | | | | | The EGL implementation used in S60/NGA for OpenVG does not return a configuration with alpha masking unless it is explicitly requested. Reviewed-by: Julian de Bhal
| | * Work around broken ATI X1600 drivers on Mac OS XGunnar Sletta2009-10-131-1/+19
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The GLSL implementation messes up return values from functions so that all our srcPixel()'s become black and several matrices are off. We don't want to rewrite the shader code to fit an "ancient" graphics card, so we simply fall back to the GL 1 engine. Reviewed-by: Trond (cherry picked from commit 33ed3d0bacddce214a43be60eb6481903e753a88)
| | * Modified QSymbianControl::setFocusSafely to set last focused controlGareth Stockwell2009-10-131-1/+2
| | | | | | | | | | | | | | | | | | to zero only if currently focused Reviewed-by: axis
| | * Mac: fix autotest for qcolordialogRichard Moe Gustavsen2009-10-131-1/+3
| | | | | | | | | | | | Rev-By: olivier
| | * crash fix on WinCE without gesture supportMaurice Kalinowski2009-10-131-3/+6
| | | | | | | | | | | | | | | | | | | | | | | | | | | dd9d8693 added some checks causing SetGestureConfig to not be initialized to 0. Thus it gets derefenced and causes crashes on all WinCE applications. Reviewed-by: denis Reviewed-by: ninerider
| | * *Completely new* EABI DEF files for Symbian OSIain2009-10-1311-20796/+18739
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Notes: - refrozen from scratch without any of the autotest exports (To spell it out: there is no BC between anything built against these DEF files, and anything built with previous versions of the DEF files) - no autotest exports in this set, so no good for autotests - it is very likely that the DEF files will be reworked again before release, meaning that anything built against these ones won't be BC with the 4.6.0 final release (so don't expect your apps built against the beta to work against the final release) - finally added a DEF file for WebKit Reviewed-by: TrustMe Conflicts: src/s60installs/eabi/QtGuiu.def src/s60installs/eabi/QtScriptu.def src/s60installs/eabi/phononu.def
| | * Temporary workaround to get WebKit to pick up DEF file from std locationIain2009-10-131-0/+13
| | | | | | | | | | | | | | | | | | | | | Tweak WebCore .pro file to get its DEF file from the same location as all the other DEF files come from Reviewed-by: TrustMe
| | * Mac: small bugfix to the mac gesture implementationRichard Moe Gustavsen2009-10-131-0/+6
| | | | | | | | | | | | | | | | | | | | | | | | Make sure that we cancel the pan gesture if the user starts to press several fingers on the trackpad, and the gesture has not yet got a chance to start Rev-By: trustme
| | * Fixed the QKeyEvent::nativeModifiers() call.axis2009-10-131-1/+1
| | | | | | | | | | | | | | | | | | | | | Earlier we returned a copy of the Qt modifiers. This patch fixes the call to return the real native Symbian one. RevBy: Sami Merila
| | * Mac: small code fix-up for the gesture commit 5 min ago.Richard Moe Gustavsen2009-10-131-2/+5
| | | | | | | | | | | | Rev-By: trust me
| | * Mac: implement gestures conforming to the new APIRichard Moe Gustavsen2009-10-136-17/+392
| | |
| | * Carbon: QApplication auto test shows bug in event dispatcherRichard Moe Gustavsen2009-10-131-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The reason for the bug is that we call _quit_ on the eventloop just _after_ posting the deffered delete event (from inside deleteLater function, ref the test where it fails, tst_qapplication.cpp:1242). And the point is, even if the loop level tells us that we _can_ delete the object in this case, the 'quit' tells us that we should not process _any_ events (until we get a call to processEvents again). So this patch makes sure that we don't call sendPostedEvents from the eventDispatcher if it is interruped. Rev-By: brad
| | * Second attempt: Fixed a few compiler warnings from QtGui for Symbian.Janne Anttila2009-10-134-4/+7
| | | | | | | | | | | | | | | | | | | | | The previous commit 9c6e466dc09813dd4a641b2cf385f35ac99346f5 included some changes which were not meant to be pushed. Reviewed-by: TrustMe
| | * Revert "Fixed a few compiler warnings from QtGui for Symbian."Janne Anttila2009-10-137-16/+11
| | | | | | | | | | | | | | | | | | Accidentally used git commit -a when should have used only git commit :( This reverts commit 9c6e466dc09813dd4a641b2cf385f35ac99346f5.
| | * Re-applying commit ee0a43fee20cc398b505eb65218ebed56dfc8f39Jocelyn Turcotte2009-10-131-2/+6
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | by Simon Hausmann Fix crash of QtScript on Mac OS X When compiling on 10.4 but running on 10.5 the flags passed to vm_map cause it to crash. For now fall back to the use of mmap() as allocator instead. Reviewed-by: Kent Hansen
| | * Remove heuristic mask from S60 cursor to match other platformsShane Kearns2009-10-131-12/+12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | S60 port was supporting QCursor(QPixmap) where the pixmap didn't have any mask or alpha channel, and generating a mask using QPixmap::heuristicMask. Now changed to draw the cursor opaque in this case, which matches the behaviour on other platforms. Also it failed horribly with shaded backgrounds e.g. draggable icons demo Task-number: QTBUG-4761 Reviewed-by: axis
| | * Merge qt-team-widgets/4.6 into origin/4.6Olivier Goffart2009-10-1311-130/+340
| | |\
| | * | Fixed a few compiler warnings from QtGui for Symbian.Janne Anttila2009-10-137-11/+16
| | | | | | | | | | | | | | | | Reviewed-by: TrustMe
| | * | Merge branch '4.6' of git@scm.dev.nokia.troll.no:qt/qt into 4.6Rhys Weatherley2009-10-132-1/+4
| | |\ \
| | | * | Doc: clarified what addDatabase() returns if the driver cannot be loadedStian Sandvik Thomassen2009-10-131-0/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | If the driver cannot be loaded, isValid() returns false. Reviewed-by: Bill King
| | | * | Fixes issue with multiple lookups to same table/fieldBill King2009-10-131-1/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Previously the renaming scheme created a new table_field alias name. If multiple references referred to the same table/field lookup, then multiples of the same alias would be generated in the query, leading the query to fail. Reviewed-by: Justin McPherson
generated by cgit v0.12 at 2025-03-07 00:35:36 (GMT)