summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Benchmarks comparing QMatrix4x4 with QTransformRhys Weatherley2009-08-201-0/+411
| | | | Reviewed-by: trustme
* Remove QGraphicsTransform::project()Rhys Weatherley2009-08-194-35/+4
| | | | | | QMatrix4x4::toTransform() now does what project() used to do. Reviewed-by: trustme
* Add projection support to QMatrix4x4::toTransform()Rhys Weatherley2009-08-192-7/+48
| | | | | | | | | | | When converting from 3D back to 2D, it is necessary to project the z component unless an orthographic projection is desired. This change adds a distanceToPlane argument that specifies the perspective projection factor to apply during the conversion. The default value of 1024 corresponds to the projection performed by QTransform::rotate(), for consistency with existing classes. Reviewed-by: trustme
* Fix comment for QGraphicsTransform::project()Rhys Weatherley2009-08-191-1/+1
| | | | | | Minor error in the documentation for the projection matrix. Reviewed-by: trustme
* Merge branch 'master' of git@scm.dev.nokia.troll.no:qt/qt into kinetic-transformRhys Weatherley2009-08-192-0/+10
|\
| * Fixes compileBill King2009-08-192-0/+10
| |
* | Re-implement QGraphicsTransform to use QMatrix4x4Rhys Weatherley2009-08-194-165/+256
|/ | | | | | | | | | QTransform-based transformations create problems when performing X and Y axis rotations because they aren't using true 3D. This change modifies QGraphicsTransform and its sub-classes to use QMatrix4x4 as the standard transformation matrix, with a project() function to project back to 2D when required. Reviewed-by: trustme
* Move math3d unit tests up one level to tests/autoRhys Weatherley2009-08-1812-74/+14
| | | | | | | | Some of the platform test infrastructure assumes that all unit tests are sub-directories under tests/auto, and the tests/auto/math3d sub-directory was confusing that infrastructure. Reviewed-by: trustme
* Make sure windows raise properly when focusedAnders Bakken2009-08-183-25/+57
| | | | | | | | | | We never used to properly raise windows with DirectFB's window management. Somehow the event happens on a window surface that doesn't really have a IDirectFBWindow associated with it but using permanentState I can establish a sibling relationship to ensure raise gets called on the right window surface. Reviewed-By: Donald <qt-info@nokia.com>
* Initialize bpl in QDirectFBPaintDevice(...)Anders Bakken2009-08-181-1/+1
| | | | Reviewed-by: Donald <qt-info@nokia.com>
* Disambiguate variable namesAnders Bakken2009-08-182-7/+7
| | | | | | | | s/lock/lockFlgs/ lock is also a function in QWSWindowSurface. Reviewed-by: Donald <qt-info@nokia.com>
* Autotest: cosmetic: change trolltech.com to qt.nokia.comThiago Macieira2009-08-181-27/+27
|
* Fixed toRawForm because the domains usually start with a dot.Thiago Macieira2009-08-181-1/+6
| | | | | | | | | | Like 9fea895d6, the series of commits ending in ff1280178 made QUrl::toAce more strict. Now it doesn't accept empty domain labels, which is exactly what a leading dot means. Interestingly, KDE 3's KURL had a long-standing hack to support the leading dot and which I broke on more than one occasion. And it had that feature exactly because of cookies.
* Autotest: update the trolltech.com URLs to nokia.com properly.Thiago Macieira2009-08-181-27/+27
|
* Autotest: Revert 96b6a3c9cd8 in tst_qnetworkcookiejarThiago Macieira2009-08-181-13/+13
| | | | | The change of URLs in Qt broke the tests that operated on URLs. This wasn't properly done, so revert to the last working values.
* Autotest: don't check Qt3Support and QtCLucene for global statics.Thiago Macieira2009-08-181-1/+9
| | | | | We won't fix any issues in those libraries anyway, so don't waste time checking.
* Remove a global static in QtTest: we can do the same with dynamicThiago Macieira2009-08-181-2/+7
| | | | allocation.
* Remove a global destructor from QtDBus.Thiago Macieira2009-08-181-4/+6
| | | | Instead, only run the unload code if we successfully loaded.
* Autotest: accept that the platform's UTF-8 codec can be buggy.Thiago Macieira2009-08-181-1/+10
| | | | | We shouldn't fail in our autotests because the platform is buggy. But note the problem. We should report the bugs upstream...
* Autotest: update uic baselines after license header update.Thiago Macieira2009-08-18100-287/+342
| | | | | The license header update in e12a03d5 changed the .ui templates, but not the baselines in all cases.
* Doc: explain that QUrl::toAce may fail if its input is not correct.Thiago Macieira2009-08-181-0/+4
|
* Autotest: unbreak test after overeager trolltech.com -> qt.nokia.com replacementThiago Macieira2009-08-181-2/+2
|
* Simplify DirectFB clippingAnders Bakken2009-08-182-25/+36
| | | | | | | | | | | | | | Instead of storing whether or not our clip is dirty and updating it in every paint operation we'll just update it whenever it changes. This is more intuitive and should be faster for the common case as well. Also make sure to implement clip(region). Prevent multiple updateClip() calls for a single clip operation by introducing a flag to check whether we recursed into the different clip() calls. Reviewed-by: Donald <qt-info@nokia.com>
* Fix bugs in DFBPE fillRect and drawTiledPixmapAnders Bakken2009-08-181-3/+2
| | | | | | Make sure to call CLIPPED_PAINT so painting with a region clip works. Reviewed-by: Donald <qt-info@nokia.com>
* Made the browser demo use the new QXmlStreamReader functionsThorbjørn Lindeijer2009-08-182-74/+36
| | | | Same as in the bookmarks example.
* Added a behaviour parameter to QXmlStreamReader::readElementTextThorbjørn Lindeijer2009-08-183-8/+108
| | | | | | | | | | | | | This makes the function a bit more useful, since previously it was only safe to use if you were sure that it would not encounter an unexpected child element, or if you would be alright with canceling the parser on such an occurrence. Now it is also possible to have it ignore any unexpected child elements, or to have it include the text found in any child elements. Task-number: 231938 Reviewed-by: mae
* Added two convenience functions to QXmlStreamReaderThorbjørn Lindeijer2009-08-187-85/+114
| | | | | | | | | | | | | | | | | QXmlStreamReader::readNextStartElement reads until the next start element within the current element, or returns false when no such element is encountered before the end element is reached. It simplifies the common case of iterating over the elements in an XML document. QXmlStreamReader::skipCurrentElement reads until the end element of the current element, skipping any child elements. This functionality was requested in two tasks, and a similar function 'readUnknownElement' was present in Qt's stream reader example. Autotest is included, example and documentation have been updated. Task-number: 238793 Reviewed-by: mae
* QGraphicsView test fixGabriel de Dietrich2009-08-181-6/+6
| | | | Reviewed-by: Olivier
* Fixed a crash when setting twice the same viewmode on a QListViewThierry Bastian2009-08-181-1/+3
| | | | autotest will detect the problem
* Add support for LIBS_PRIVATE to the VCProject generator too.Thiago Macieira2009-08-181-1/+3
| | | | Reviewed-by: Marius Storm-Olsen
* Fixed QRasterPaintEngine::drawImage() for 1x1 source rectangles.Kim Motoyoshi Kalland2009-08-181-2/+34
| | | | | | | | | | | | | | | | | | | | | QRasterPaintEngine::drawImage() used an optimized code path when the source rectangle was 1x1. It would sample the source image at the source rectangle's top-left corner and use the sampled color to fill the entire target rectangle. There were two bugs, however: 1) The sampled color was assumed to be non-premultiplied, so you could end up with premultiplying the color twice. This was fixed by avoiding premultiplying a second time if the source image has a premultiplied format. 2) Since the source rectangle is a QRectF, it could easily cross pixel boundaries even if it's 1x1. In this case, it is not correct to fill the target rectangle with a single color. This was fixed by checking if the entire source rectangle is contained in a single pixel before taking the optimized code path. Task-number: 256950 Reviewed-by: Trond
* Merge branch 'master' of git@scm.dev.nokia.troll.no:qt/qtDavid Boddie2009-08-1873-901/+655
|\
| * Make sure to update pos QDFBCursor::moveAnders Bakken2009-08-181-0/+1
| | | | | | | | | | | | | | | | We don't need to chain to the base class since the QDirectFBScreenCursor always is rendered by hardware but we should update the pos variable in case people ask where the cursor is. Reviewed-by: TrustMe
| * mingw: make it possible to build without rtti supportThierry Bastian2009-08-181-0/+1
| |
| * add QGraphicsObject Q_INTERFACES macroThierry Bastian2009-08-181-0/+1
| |
| * Compile fix with namespaceshjk2009-08-181-0/+2
| |
| * Animations: update the documentationThierry Bastian2009-08-181-7/+0
| | | | | | | | Removed the reference to the QAnimationState that doesn't exist anymore
| * Make TestLib compatible with mingw when gui is not usedThierry Bastian2009-08-183-2/+7
| | | | | | | | | | | | | | | | The problem is that headers in QTestLib define functions for mouse and key events. Those are causing link errors on all autotests that do "QT -= gui" with mingw (only debug seems to be affected). Reviewed-by: jasplin
| * Fixed bug when rendering long lines of text without XRender.Samuel Rødal2009-08-181-4/+12
| | | | | | | | | | | | | | XRectangle coordinates need to be clipped to the short integer range. Task-number: 250137 Reviewed-by: Trond
| * Fix autotest: on windows, timers are not accurate enoughThierry Bastian2009-08-181-2/+4
| |
| * Autotest: Be very pedant about the tests. There's at least one compilerThiago Macieira2009-08-181-4/+33
| | | | | | | | where it breaks, so let's find out where.
| * Fix autotest: removed useless debug tracesThierry Bastian2009-08-181-1/+0
| |
| * QItemSelectionModel: hasSelection can return true when no selectionThierry Bastian2009-08-181-11/+11
| | | | | | | | | | | | we needed to finalize the selection when rows are removed Reviewed-by: ogoffart
| * QHeaderView: assertion when modifying the root iem of a treewidgetThierry Bastian2009-08-181-1/+2
| | | | | | | | | | | | | | dataChanged is emitted with invalid model index. We needed to handle that. Reviewed-by: ogoffart
| * Fix build failure on Mac OS 10.5Richard Moe Gustavsen2009-08-181-2/+4
| | | | | | | | Gestures only awailable on 10.6
| * Added support for gestures on Mac OS X CarbonRichard Moe Gustavsen2009-08-183-7/+76
| |
| * Speed up midpoint lines starting far outside the device bounds.Gunnar Sletta2009-08-181-0/+14
| | | | | | | | Reviewed-By: Samuel
| * tst_QTableView::task259308_scrollVerticalHeaderSwappedSections fails.Gabriel de Dietrich2009-08-181-0/+1
| | | | | | | | | | | | | | | | The QTableView scrolled down after the sections swap. Setting the current index to the topmost visual element fixes the issue. Reviewed-by: Thierry
| * As of Qt 4.6, the minimum OS X version is 10.4.Morten Sorvig2009-08-181-1/+1
| |
| * Fixed a possible crash in QTextCursorThierry Bastian2009-08-181-1/+2
| |
generated by cgit v0.12 at 2024-11-15 03:24:45 (GMT)