summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Add license headersPaul Olav Tvete2009-08-313-0/+123
| | | | Reviewed-by: trustme
* Fix QSslCertificate::alternateSubjectNames() memory leakRaul Metsma2009-08-313-1/+4
| | | | | | | Task-number: 258964 Merge-request: 1382 Reviewed-by: Peter Hartmann <peter.hartmann@trolltech.com>
* Preemptively parse the necessary attributes in QSvgAttributes.Ariya Hidayat2009-08-311-57/+195
| | | | | | | | | | | | | | | | | | | | | | | | | | | | Previously, retrieving an attribute value requires a two-stage lookup, first in the XML attributes and (if it is found there) in the CSS attributes. Both look-ups requires a number of iterations and comparisons, which have to be carried out for every value look-up. This patch changes it so that iterations and comparisons need to be done only once, namely at the beginning. This requires us to store several SVG attributes needed by the parsing routine, but since they are just QStringRefs, the increase in the heap usage is really minimal and even not reported by Valgrind's Massif. Also, now we don't need to hold the XML attributes anymore. The loading of tiger.svg (tests/benchmarks/qsvgrenderer) is reduced from 101.2 millions instructions to 96.5. The biggest gain is however obvious from the time spent in QSvgAttributes::parseStyle(QSvgNode*, QSvgAttributes, QSvgHandler*) which goes down from 16.7 millions instructions to 6.9 millions, i.e. 2.4x faster. Even with the new extra overhead in the QSvgAttributes constructor, QSvgAttributes::parseStyle(QSvgNode*, QXmlStreamAttributes, QSvgHandler*) goes down from 23.5 millions instructions to 18.4 millions, i.e. 1.3x faster. Reviewed-by: Kim
* fix crash due to double deletionKent Hansen2009-08-312-1/+22
| | | | Needed due to commit 3636e666528b72de79f8c7012690bb9e279f0863
* Merge branch '4.6' of git@scm.dev.nokia.troll.no:qt/qt into 4.6Martin Smith2009-08-3127-180/+1146
|\
| * don't disable the JIT for qtscriptKent Hansen2009-08-311-2/+0
| | | | | | | | AKA, enable it on platforms where it's supported.
| * make sure hasInstance() is always invoked for custom script objectsKent Hansen2009-08-311-1/+1
| | | | | | | | | | | | | | Need to specify the OverridesHasInstance flag, otherwise the JIT will do the standard hasInstance implementation instead of calling our virtual function. This makes the QScriptClass::extension() autotest pass.
| * implement functionExit() callbacks on the JITKent Hansen2009-08-313-20/+42
| | | | | | | | | | | | The return value is not reported (we need a separate callback for that, just like in the interpreter), but that isn't important to get our tools (i.e. the debugger) working.
| * No need for the namespaced version in QSvgAttributes::value().Ariya Hidayat2009-08-311-18/+3
| | | | | | | | | | | | | | We do not really use this function with proper namespace URI, so let us get rid of the function. Reviewed-by: Kim
| * Merge branch '4.6'Thiago Macieira2009-08-3136-242/+1259
| |\
| | * Merge branch '4.5' into 4.6Thiago Macieira2009-08-318-103/+177
| | |\ | | | | | | | | | | | | | | | | Conflicts: tests/auto/xmlpatternsxqts/lib/tests/XMLWriterTest.cpp
| | | * Autotest: add some D-Bus tests sending complex structs and lists of themThiago Macieira2009-08-312-0/+53
| | | |
| | | * Fix license headers.Jason McDonald2009-08-315-102/+123
| | | | | | | | | | | | | | | | Reviewed-by: Trust Me
| | | * Merge branch '4.5' of git@scm.dev.nokia.troll.no:qt/qt into 4.5Keith Isdale2009-08-31694-3089/+24515
| | | |\
| | | * | Correct grammatical in documentation for QNetworkDiskCache::expire()Keith Isdale2009-08-311-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The text: When the current size of the cache is greater then the maximumCacheSize() should read When the current size of the cache is greater than the maximumCacheSize() Task-number: 260496 Reviewed-by: Zheng Liu (George)
| | * | | Add an opt-in QWeakPointer::operator->.Thiago Macieira2009-08-311-0/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | To use this, you must add #define QWEAKPOINTER_ENABLE_ARROW to your .cpp before #including any headers. Be careful about adding this to .h files. Reviewed-by: Olivier Goffart
| | * | | Benchmark test for QSvgRenderer.Ariya Hidayat2009-08-314-0/+848
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Right now mostly testing the constructor and the parser. The only SVG to be tested so far is tiger.svg, i.e. the famous Tiger head converted from tiger.eps found in GPL-ed Ghostscript. Reviewed-by: Kim
| | * | | Merge branch 'master' into 4.6Janne Anttila2009-08-314-14/+58
| | |\ \ \
| | | * \ \ Merge branch 'master' of git@scm.dev.nokia.troll.no:qt/qt-s60-publicJanne Anttila2009-08-281-1/+1
| | | |\ \ \
| | | * | | | Updated docs to reflect current 'createpackage' and 'make sisx' syntax.Janne Anttila2009-08-284-14/+58
| | | | | | | | | | | | | | | | | | | | | | | | | | | | Reviewed-by: Miikka Heikkinen
| | * | | | | Fixed createpackage to use lowercase target to be Linux compliant.Janne Anttila2009-08-311-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | All S60 SDK paths in Linux are lowercase. Reviewed-by: TrustMe
| | * | | | | Trivial change to comment.Janne Anttila2009-08-311-0/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | Reviewed-by: TrustMe
| | * | | | | Removed hard-coded sqlite3 deployment statement - used PKG conditions.Janne Anttila2009-08-311-11/+7
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Before this change Qt did not install to device which had sqlite3 already in ROM. Since we now install the sqlite3 only if it is not on any drive, the installation works also with preinstalled sqlite. Even better appraoch would be to use embedded SIS file for sqlite, since that would allow sqlite3 upgrade without upgrading the whole Qt. However we yet don't have such SIS, but once we get one from Symbian this needs to be further changed. Task-number: 258858 Reviewed-by: Miikka Heikkinen
| * | | | | | Update the qmake autotest: re MSVC test failure findMocs and findDepsKeith Isdale2009-08-312-10/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When MSVC is used the default DESTDIR makes use a release/debug suffix. For these tests DESTDIR needs to be "."/ Minor code cleanup was done whilst correcting these effected .pro files Reviewed-by: Lincoln Ramsay
* | | | | | | doc: Fixed several qdoc errors.Martin Smith2009-08-314-13/+55
| |/ / / / / |/| | | | |
* | | | | | Merge branch '4.6' of git@scm.dev.troll.no:qt/qt into 4.6Janne Anttila2009-08-312-6/+19
|\ \ \ \ \ \
| * | | | | | Fixes compile for my last change.Bill King2009-08-311-1/+1
| | | | | | |
| * | | | | | Adds ability to open sqlite databases readonly.Bill King2009-08-312-6/+19
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Changes opening of sqlite to open_v2, and adds a new connection option. Reviewed-by: Justin McPherson
* | | | | | | Added QContextMenuEvent sending to widget in Symbian.Janne Anttila2009-08-311-1/+2
|/ / / / / / | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The event construction without sending does nothing, sending constructed event possibly dissappeared due to bad merge or something. So here it comes again... Task-number: 260222 Reviewed-by: TrustMe
* | | | | | Added doc link to point to Assistant doc table of collection file values.Peter Yard2009-08-311-0/+8
| | | | | | | | | | | | | | | | | | | | | | | | Fix for task 259252.
* | | | | | Move QGLShareContextScope to qgl_p.h so other things can use it.Rhys Weatherley2009-08-312-37/+40
| | | | | | | | | | | | | | | | | | | | | | | | Reviewed-by: trustme
* | | | | | Merge branch '4.6' of git@scm.dev.nokia.troll.no:qt/qt into 4.6Rhys Weatherley2009-08-311-11/+11
|\ \ \ \ \ \
| * | | | | | Fixes race condition on reopening an ODBC connection.Bill King2009-08-311-11/+11
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Original code was keeping it's own (out of date) copy of the handles. It now includes a pointer to the driver private, and asks that what the handles are instead.
* | | | | | | Remove QGLShader from its QGLShaderProgram when it is destroyedRhys Weatherley2009-08-312-4/+23
|/ / / / / / | | | | | | | | | | | | | | | | | | Reviewed-by: Samuel
* | | | | | Destroy EGL window surface when direct VG surface is deletedRhys Weatherley2009-08-301-0/+6
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | OpenVG window surfaces were leaking because the QVGEGLWindowSurfaceDirect destructor was previously not cleaning up the EGLSurface object. Reviewed-by: trustme
* | | | | | trivial: make moc compile againHarald Fernengel2009-08-281-0/+2
| | | | | |
* | | | | | Warning --Alessandro Portale2009-08-281-0/+2
| | | | | |
* | | | | | Added a default theme for the simulated S60 styleAlessandro Portale2009-08-284-3/+3
| | | | | | | | | | | | | | | | | | | | | | | | Reviewed-By: Trustme
* | | | | | QMetaObjectPrivate: refactor internal disconnect to avoid duplicate codeLeonardo Sobral Cunha2009-08-282-49/+47
| | | | | | | | | | | | | | | | | | | | | | | | Reviewed-by: ogoffart
* | | | | | Removing an obviously obsolete LIBS includion workaround.Alessandro Portale2009-08-281-2/+0
|/ / / / / | | | | | | | | | | | | | | | Reviewed-By: Trustme
* | | | | Merge branch '4.6'Thiago Macieira2009-08-2870-728/+1438
|\ \ \ \ \
| * | | | | Mark the QSharedPointer with Qt::Uninitialized explicitThiago Macieira2009-08-282-1/+13
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | That ensures we don't create ambiguities. QSharedPointer shouldn't have any implicit constructors. Suggested-By: Olivier Goffart
| * | | | | Using QFocusFrame to visualize the focus.Alessandro Portale2009-08-283-118/+80
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When using the keypad navigation, the focus state of a widget is supposed to be well visible. So it makes much sense to use QFocusFrame, which is able to paint a focus marker _around_ the widget. This change should fix the issue that with some S60 themes, the focus was invisible. Also, a lot of code is removed by this change. Task-number: 256470 Reviewed-by: Sami Merila modified: src/gui/styles/qs60style.cpp modified: src/gui/styles/qs60style.h modified: src/gui/styles/qs60style_p.h
| * | | | | Don't call qFuzzyIsNull on every single frame in QGraphicsOpacityEffect.Bjørn Erik Nilsen2009-08-282-3/+10
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | We only have to do it when the opacity changes (in ::setOpacity). Also, we cannot do the cut-off when the effect has an opacity mask set.
| * | | | | Add an autotest to check QPainter rendering to a QGLPixelBuffer worksTom Cooksey2009-08-281-0/+33
| | | | | | | | | | | | | | | | | | | | | | | | Reviewed-by: Trustme
| * | | | | Add support for setting an opacity mask on QGraphicsOpacityEffect.Bjørn Erik Nilsen2009-08-284-6/+106
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Opacity masks enable you to make portions of an element transparent. More information about opacity masks here: http://msdn.microsoft.com/en-us/library/ms743320.aspx Reviewed-by: Samuel
| * | | | | Fix wrong transform when drawing graphics item effect source.Bjørn Erik Nilsen2009-08-281-4/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | We are only interested in the difference between the original transform and the transform set on the painter we're about to use.
| * | | | | Merge branch '4.6' of git@scm.dev.nokia.troll.no:qt/qt into 4.6Gunnar Sletta2009-08-28646-1773/+24200
| |\ \ \ \ \
| | * | | | | Make it possible for autotests to check if JIT is enabled or notKent Hansen2009-08-283-10/+51
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Currently there are some differences in behavior and availability of information between the interpreter and the JIT. This is now documented as expected failures in the relevant autotests.
| | * | | | | Re-enabled the QPixmap cleanup hooks in QPixmap destructor.Trond Kjernåsen2009-08-281-0/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | Reviewed-by: Gunnar