summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Merge branch '4.6' of git@scm.dev.nokia.troll.no:qt/qt-graphics-team into 4.6Gunnar Sletta2009-11-0913-352/+554
|\
| * Fixed handling of the "..." button in QPrintDialog for X11.Trond Kjernåsen2009-11-091-5/+4
| | | | | | | | | | Task-number: QTBUG-5547 Reviewed-by: Kim
| * Fixed bug where JPEG scaling resulted in an uninitialised pixel column.Kim Motoyoshi Kalland2009-11-091-1/+4
| | | | | | | | | | | | | | | | | | | | | | Loading a JPEG image after calling QImageReader::setScaledSize() could cause a 1 pixel wide column of uninitialised pixels at the far right of the image. The bug was caused by rounding in the scaling algorithm. Fixed by copying the last calculated pixel all the way to the end of the row. Task-number: QTBUG-3282 Reviewed-by: Trond
| * Replaced tabs with spaces in scale() in qjpeghandler.cpp.Kim Motoyoshi Kalland2009-11-091-229/+238
| | | | | | | | | | | | | | Replaced tabs with spaces and Qt-ified code formatting in QImageSmoothScaler::scale(). Reviewed-by: Trond
| * Fixed loading of 8-bit grayscale TIFF files.Samuel Rødal2009-11-095-63/+118
| | | | | | | | | | | | | | | | | | | | | | The PHOTOMETRIC_MINISBLACK and PHOTOMETRIC_MINISWHITE settings also apply to 8-bit grayscale, so we need to check the bit depth of the image as well. For convenience we also try to write the images using PHOTOMETRIC_MINISBLACK and PHOTOMETRIC_MINISWHITE as opposed to PHOTOMETRIC_PALETTE when we detect that the color table is grayscale. Task-number: QTBUG-5459 Reviewed-by: Benjamin Poulain
| * Add a simpler FBO autotest which just uses glClear()Tom Cooksey2009-11-091-0/+31
| | | | | | | | If this autotest doesn't pass, FBOs are really broken.
| * Mark both the new and original EGL contexts as sharingRhys Weatherley2009-11-094-0/+7
| | | | | | | | | | | | | | If we succeed in creating a shared context, then mark the original as shared as well. Reviewed-by: Sarah Smith
| * Detect context sharing on EGL systemsRhys Weatherley2009-11-095-0/+9
| | | | | | | | | | | | | | | | Context sharing was enabled on EGL systems, but QGLContext::isSharing() was still returning false because the flag in QGLContextPrivate was not updated. Reviewed-by: Sarah Smith
| * Set frameless window hints on QGL test windows under QWSRhys Weatherley2009-11-091-0/+15
| | | | | | | | | | | | | | | | | | | | | | | | | | | | QWS combines the window title bar and the top-level widget into one window surface, which causes problems when the top-level widget is a QGLWidget: the raw GL surface contents are not the same as on other platforms where only the top-level widget contents are included in the window surface. This change sets Qt::FramelessWindowHint on QWS systems on widgets where we need to read back the top-level widget for comparison against a reference image. Reviewed-by: Sarah Smith
| * Make qgl autotests more stable & passing in test farmTom Cooksey2009-11-061-6/+18
| | | | | | | | | | | | | | | | | | There were problems with false-failures due to the test farm's GL implementation having off-by-one pixel errors. To fix this, we don't compare every pixel but rather sample pixels in a grid pattern which avoids boundries. Reviewed-By: Samuel
| * Enable & fix qgl autotests on 16-bpp systemsTom Cooksey2009-11-061-49/+111
| | | | | | | | | | | | | | | | | | Introduce a fuzzy pixel and image compare which changes it's allowed fuzz based on the system's color depth. If the compared images are different, the autotests will also save the two images to the current directory for comparison. Reviewed-By: Samuel
* | Fixed dropshadow raster implementation since rewriteGunnar Sletta2009-11-091-4/+9
| | | | | | | | Reviewed-by: Samuel
* | Fixed up qttracereplay and trace graphicsystem a bit.Gunnar Sletta2009-11-093-10/+31
| | | | | | | | | | | | | | | | | | | | | | - Added "qttrace" to binary file, making it possible to do a sanity verification of the input file in qttracereplay - Sanity check input binary file, both for existance and for content in qttracereplay - Console app on windows and mac - Added helpful output for -h and --help command line options Reviewed-by: Samuel
* | Pixelalign circles in HTML bullet listsGunnar Sletta2009-11-091-1/+1
| | | | | | | | Reviewed-by: Eskil
* | Fixed inconsistent coloring of bullets, circles and square listsGunnar Sletta2009-11-091-1/+1
|/ | | | | | | Also removed a redundant setBrush(), as it will be followed by painter->restore() a few lines below. Reviewed-by: Eskil
* Merge branch '4.6' of git@scm.dev.nokia.troll.no:qt/qt into 4.6Gunnar Sletta2009-11-061-0/+1
|\
| * Merge branch '4.6' of git@scm.dev.nokia.troll.no:qt/qt into 4.6Jason McDonald2009-11-0661-135/+824
| |\
| * | Fix error when configuring on Unix using an evaluation license.Jason McDonald2009-11-061-0/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | If you run configure on Unix more than once using an eval license (or -DQT_EVAL), configure outputs an error and fails to overwrite src/corelib/global/qconfig_eval.cpp. This is corrected simply by removing any existing instance of the file before re-creating it. Reviewed-by: Trust Me
* | | Fixed g++ compiler warnings in qpen.cpp.Samuel Rødal2009-11-061-1/+1
| | |
* | | Fix fuzzy aliased rendering on GLES2Tom Cooksey2009-11-062-2/+13
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The GL2 paint engine adds a (0.49,0.49) pixel offset when doing aliased rendering. But this assumed if it was doing aliased rendering then multisampling was disabled. On GLES, multisampling is always enabled if the surface has it enabled. So on GLES, we never add the offset if the surface is multisampled. Reviewed-By: Gunnar
* | | Merge branch '4.6' of git@scm.dev.nokia.troll.no:qt/qt-graphics-team into 4.6Gunnar Sletta2009-11-062-15/+41
|\ \ \
| * | | Fix QPainter::setPen(pen with color but no style) on non-extended.David Faure2009-11-062-15/+41
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | It was working with a QImage but not with a QPixmap, which is obviously a bug. In addition, it broke WebCore::GraphicsContext::setPlatformStrokeColor which does exactly what I put in the unittest: get pen, set color, set pen. This commit fixes the wrong color in the underline of the links in http://www.davidfaure.fr/kde/link_underline_color.html in QtWebkit. Merge-request: 1995 Reviewed-by: Samuel Rødal <sroedal@trolltech.com>
* | | | Merge branch '4.6' of git@scm.dev.nokia.troll.no:qt/qt into 4.6Gunnar Sletta2009-11-0663-166/+945
|\ \ \ \ | |/ / / |/| | / | | |/ | |/|
| * | Merge commit 'origin/4.6' into large-file-supportJoão Abecasis2009-11-0616-133/+187
| |\ \ | | |/
| | * my change log for 4.6Thierry Bastian2009-11-061-9/+61
| | |
| | * Support composition mode source for drawing in dfbAnders Bakken2009-11-061-19/+52
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Qt's/DirectFB's composition modes are not compatible for drawing operations. We only support SourceOver and handle whether or not to blend colors using DSDRAW_BLEND based on the alpha/opacity value. If drawing with Source and an opaque color we can still handle the operation using DirectFB. This has major impact since WebKit fills the background of web pages using Source. Reviewed-by: Donald Carr <donald.carr@nokia.com>
| | * Reset composition mode after drawing in dfbAnders Bakken2009-11-061-3/+8
| | | | | | | | | | | | | | | | | | | | | | | | When drawing primitives in DirectFB we always set composition mode to NONE and manually manage whether or not to blend. This patch makes sure we will reset the composition mode afterwards. Reviewed-by: Donald Carr <donald.carr@nokia.com>
| * | Merge commit 'origin/4.6' into large-file-supportJoão Abecasis2009-11-052363-58757/+101885
| |\ \
| * | | Switch large file test to using Unbuffered modeJoão Abecasis2009-11-051-2/+2
| | | | | | | | | | | | | | | | | | | | ... so we test the file engine directly and detect file-system errors earlier.
| * | | Updating documentation to reflect actual behaviorJoão Abecasis2009-11-041-9/+11
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | QFile::seek is supposed to work on Windows CE with FILE* streams. size() returns 0 on errors and sequential files. Probably on empty files as well ;-) Reviewed-by: Olivier Goffart
| * | | Further fixes to file size handling on Windows with fd and FILE*João Abecasis2009-11-041-17/+26
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | filelength is not available on Windows CE instead, we must fallback to fseek/ftell as was being done previously. Still on Windows CE, we still don't report the file size for file descriptors, but we also won't set a random error string. Changed qt_error_string calls to use errno when errors come from CRT functions. Also, if we're using filelength on FILE* streams, there's no reason not to use it for file descriptors, instead of requesting a native handle. Reviewed-by: Olivier Goffart
| * | | Turns out 64-bit fseek/ftell are not available on VS 2003/2002...João Abecasis2009-10-292-2/+9
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Not when linking dynamically to the CRT (/MT). So we can't rely on them. The declarations for those are also not on the standard headers. Reverts "(MSVC 2002/2003) Use 64-bit versions of ftell and fseek", fixes return type of QT_FTELL and skips known failures on large-file test case.
| * | | Don't try to mmap past EOFJoão Abecasis2009-10-281-0/+6
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | On Mac OS, mmap would succeed, returning a valid pointer, but trying to read from it would result in a SIGBUS. By adding this check we commit to a safe cross-platform behavior users can depend on. Reviewed-by: Thiago Macieira
| * | | (MSVC 2002/2003) Use 64-bit versions of ftell and fseekJoão Abecasis2009-10-271-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | MSDN documents these as being available since Windows 95. Reviewed-by: Marius Storm-Olsen
| * | | Define QT_OPEN_LARGEFILE on Symbian + WinCEJoão Abecasis2009-10-272-0/+2
| | | | | | | | | | | | | | | | Reviewed-by: Thiago Macieira
| * | | Fix silly error in bit fiddlingJoão Abecasis2009-10-261-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | That's what I get for not having the brain on when accepting suggestions... Reviewed-by: Thiago Macieira
| * | | Extending QFile::size test to cover files opened with fd and FILE*João Abecasis2009-10-231-8/+43
| | | | | | | | | | | | | | | | | | | | | | | | | | | | Also changed tested type from int to qint64, so we'll be able to see clipping issues, although there are no large files in this test, yet. Reviewed-by: Markus Goetz
| * | | Fix QFile::isSequential on WindowsJoão Abecasis2009-10-231-19/+10
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When not using native HANDLEs, the return of isSequential was hardcoded to true for files with a fd, and for the standard FILE* streams stdin, stdout and stderr; false for all other FILE* streams. We now use the native GetFileType call for all files by obtaining a native handle where required. We also treat files of type FILE_TYPE_CHAR as sequential, as is the case for the standard streams in console applications. When standard streams are redirected to/from files, GetFileType will return FILE_TYPE_DISK for them and they won't be considered sequential. This is alright since in this mode they behave like regular files and QFile::seek() will work for random offsets. Reviewed-by: Marius Storm-Olsen
| * | | Fix the LargeFile test for WindowsJoão Abecasis2009-10-231-10/+15
| | | | | | | | | | | | | | | | | | | | | | | | The test assumed fileName was stable, but it is documented behaviour that this can be reset when a file descriptor or FILE* stream is associated with a QFile. This was the case on Windows.
| * | | On Windows, report a 0 file size for streams and other funny filesJoão Abecasis2009-10-231-5/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Obtain file size directly, instead of relying on fseek/ftell and messing with the file position. Also changed the return value on errors to 0. This is mostly relevant on streams and pipes, where we used to return whatever error value ftell returned (usually -1). This change also makes the return value consistent with what is returned on Unix platforms and what we document for Windows CE. Nevertheless, documentation of this and related issues is misleading and needs to be updated. Reviewed-by: Markus Goetz
| * | | Check for non-zero return from fseekJoão Abecasis2009-10-231-4/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | While POSIX specifies a -1 return on error, on Windows only non-zero is documented for error conditions. All platforms agree that zero is returned on success so we check for that instead. Reviewed-by: Markus Goetz
| * | | Windows doesn't #define STD{IN,OUT,ERR}_FILENOJoão Abecasis2009-10-231-0/+12
| | | | | | | | | | | | | | | | Reviewed-by: Markus Goetz
| * | | Adding a test case for large file supportJoão Abecasis2009-10-213-1/+497
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The test case creates a (tentatively sparse) very large file with scattered data and uses it to test various aspects of large file support in QFile. Reviewed-by: Thiago Macieira
| * | | Get file position when attaching an open file descriptor to QFileJoão Abecasis2009-10-212-1/+61
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This was already being done when attaching to FILE* streams. Doing the same here makes the API consistent and more usable. Namely, one can use QFile::pos() to obtain the file position. Test case verifies this doesn't break support for sequential files. More thorough test case included in large file support test. Reviewed-by: Thiago Macieira
| * | | Fixes some issues with large files in 32-bit systemsJoão Abecasis2009-10-212-2/+5
| | | | | | | | | | | | | | | | Reviewed-by: Thiago Macieira
| * | | (Windows) Don't create a file mapping for each view that is mappedJoão Abecasis2009-10-213-62/+34
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Creating a file mapping for each view mapped to memory is sub-optimal and slow. With this change, a single mapping is created and reused when mapping subsequent views of the file. The handle returned from CreateFileForMapping (used in WinCE 5) is now discarded, since the kernel manages it automatically with the file mapping. This simplifies use of the deprecated map API. Reviewed-by: Maurice Kalinowski Reviewed-by: Marius Storm-Olsen
| * | | Fix 32/64-bit issues with QFile::map/unmap() on *nix systemsJoão Abecasis2009-10-212-11/+22
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | We would previously silently ignore overflows in 32-bit systems and not properly support 64-bit offsets in systems that support it because of integer overflow. There was also a problem that could prevent unmap from succeeding, because we were passing the wrong length argument. Task-number: QT-1594 Reviewed-by: Thiago Macieira
| * | | Introducing QT_MMAPJoão Abecasis2009-10-2151-0/+81
| | | | | | | | | | | | | | | | | | | | | | | | | | | | In platforms with Large File Support, we use mmap64, to be able to handle offsets (and therefore files) larger than 2Gb. Reviewed-by: Thiago Macieira
* | | | Merge branch '4.6' of git@scm.dev.nokia.troll.no:qt/qt-graphics-team into 4.6Gunnar Sletta2009-11-0612-70/+210
|\ \ \ \
| * | | | CompileEskil Abrahamsen Blomfeldt2009-11-061-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Missing comma in qfontdatabase.cpp Reviewed-by: Gunnar
generated by cgit v0.12 at 2025-08-26 12:07:05 (GMT)