| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
STL support hasn't had binary compatibility effects for the entire
lifetime of Qt 4, so it should never have been there. It is a legacy
thing I am now correcting.
When inspecting a plugin, remove the "no-stl" from its loaded build
key. That indicates a pre-4.6 build, since now Qt no longer adds it to
its own build keys.
Note that you have to remove the 4.6 plugin cache from
$HOME/.config/Trolltech.conf or the registry for this change to work
(if you compile Qt with -no-stl). Otherwise, plugins that have already
been scanned will fail to load.
Reviewed-by: Bradley T. Hughes
|
|
|
|
|
|
|
|
| |
This makes Qt work on QNX 6.4.
* no q3support, no phonon
* no QSharedMemory, no QSystemSemaphore, no QProcess
Reviewed-By: Robert Griebl
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This makes Qt work on VxWorks 6.6+ in native (kernel) mode.
* compiles with the WindRiver GNU toolchain (Linux only)
* works with QWS (tested with the VNC driver only)
* tested on PPC hardware and the x86 VxWorks simulator
* no q3support, no phonon, no webkit
* no QSharedMemory, no QSystemSemaphore, no QProcess
* only one QApplication instance (flat address space)
* filesystem support depends heavily on the quality of the native driver
* QLibrary is just a dummy to make plugins work at all
* qmake transparently creates VxWorks munching rules for static ctors
* made auto-test cope with missing OS features
A special note regarding the Q_FOREACH patch for dcc:
when calling foreach(a,c) with c being a function returning a container,
the compiler would generate 5 references to some labels (.LXXXX), which
are not there (so the linker complains in the end).
Seems like dcc doesn't really like the 'true ? 0 : <function call to get type>'
statement
Reviewed-By: Harald Fernengel
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
QtCore compiled, but of course that means very little since most of
QtConcurrent's problems are in the template code, which isn't
instantiated inside QtCore. Examples and tests all failed to build...
I'm almost done getting it to work with that compiler, but it will
take me a little more time. I'm refactoring a bit of the QtConcurrent
code, so that things compile more smoothly, using partial template
specialisation.
To be on the safe side, re-disable for xlC 7 too.
|
|
|
|
|
|
|
|
|
| |
* removed the vr41xx, yopy and sl5000 drivers (old PDAs)
* removed the bus mouse driver (ISA days should be over)
* renamed the LinuxIS mouse driver to LinuxInput (consistency with the key driver)
* unified the LinuxInput mouse and key driver I/O handling
Reviewed-By: Paul
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| | |
Using qglobal.h and checking the compiler version with the
preprocessor has the side-effect that moc won't generate proper code
since it doesn't know about the compiler version.
Enable both modules under Sun CC 5.9 and IBM xlC 7.0.
|
| | |
|
| |
| |
| |
| |
| |
| |
| | |
'y' rather then 'yes'
Merge-request: 945
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
|
|/
|
|
| |
Reviewed-by: Lincoln Ramsay
|
|
|
|
|
|
|
| |
It was reported to be auto-detected, but wasn't. Apparently, AIX 6's
X11 doesn't have this.
Reviewed-By: Denis Dzyubenko <denis.dzyubenko@nokia.com>
|
|
|
|
|
|
|
|
|
| |
-developer-build, part 1.
Adds QT_CONFIG+=private_tests to qconfig.pri when Qt is configured
with -developer-build.
Reviewed-by: Michael Goddard
|
|
|
|
|
|
| |
.../configure: line 4462: [: missing `]'
Reviewed-by: Jason McDonald
|
|\ |
|
| |
| |
| |
| |
| |
| |
| | |
ShivaVG runs on top of OpenGL. Include the QMAKE_*_OPENGL flags
on the link line if it is needed by the OpenVG engine.
Reviewed-by: trustme
|
| |
| |
| |
| |
| |
| |
| |
| | |
We have said in the past that 10.3 has been dropped for 4.6, but we
actually have to do the work and make sure that it happens. This is the
first step.
Reviewed by: Morten Sørvig
|
| |
| |
| |
| |
| | |
This change also moves the EGL support classes from QtOpenGL to QtGui
so they can be shared between OpenGL and OpenVG.
|
|/
|
|
| |
Reviewed-by: Bill King
|
|\
| |
| |
| |
| |
| |
| | |
Conflicts:
configure.exe
tests/auto/qpainter/tst_qpainter.cpp
translations/qt_ru.ts
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
it simply breaks too much (embedded builds in particular).
will come up with something better in master.
note however, that the .qm files do NOT return to the repository. the
release manager needs to run "make qm" before packaging, just like
Qt-from-git users who want translations do (i.e., almost nobody).
Reviewed-by: Jason McDonald
|
|\ \
| |/ |
|
| |
| |
| |
| | |
Reviewed-by: Trust Me
|
| |
| |
| |
| | |
Reviewed-by: Trust Me
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
so "qmake -r && make" worked fine ... but configure does not simply run
"qmake -r".
also, now we have to build lrelease even when we don't build the other
tools.
Reviewed-by: Jason McDonald
Reviewed-by: mariusSO
|
| |
| |
| |
| | |
Reviewed-by: TrustMe
|
| |
| |
| |
| |
| |
| |
| | |
This gets rid of a few warnings when shadow-building Qt, since the
symlinks are no longer created in the source tree.
Reviewed-by: Paul Olav Tvete <paul.tvete@nokia.com>
|
| |
| |
| |
| |
| |
| |
| |
| | |
This makes sure that we only detect and use libraries that are actually
in the SDK.
Task-number: 109757
Reviewed-by: Richard Moe Gustavsen
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Some older versions of the Bourne shell (/bin/sh) don't
support the ! operator for negating the exit code of a pipeline.
In particular, this is true on certain (all?) Solaris 9 platforms.
The patch replaces all occurrences of the ! operator with equivalent
(but slightly less elegant) expressions that test the $? variable instead.
Reviewed-by: msorvig
|
| |
| |
| |
| |
| |
| | |
If unixodbc detection fails, try and link/detect iodbc libraries and link against them.
Reviewed-by: Lincoln Ramsay
|
|\ \
| |/
| |
| |
| | |
Conflicts:
tests/auto/qtreeview/tst_qtreeview.cpp
|
| |
| |
| |
| |
| |
| |
| |
| | |
Instead of the multiple character-string replacements, just check for
the discrete -arch values that we want to find. This makes the code
clearer and should reduce the chance of subtle errors.
Reviewed-by: Jason McDonald
|
|\ \
| |/ |
|
| |
| |
| |
| | |
Reviewed-by:cduclos
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The previous "fix" caused the effect of the line before the change to be
discarded, which breaks the mac binary package builds. When reviewing
changes, please read the lines of context given in the patch - they are
included for a reason.
Acked-by: Lincoln Ramsay
Acked-by: Rohan McGovern
|
|\ \
| |/ |
|
| |
| |
| |
| |
| |
| |
| |
| | |
The reason is the tha config.test ssl failed building. And the reason for
that was a bad makefile flag (i386_64).
Task-number: 253887
Reviewed-by: Trenton Schulz
|
|\ \
| |/
| |
| |
| |
| | |
Conflicts:
src/gui/painting/qbackingstore.cpp
src/gui/painting/qwindowsurface_raster.cpp
|
| |
| |
| |
| |
| |
| |
| | |
STRING="${STRING/x86/i386}" actually replaces whole words only,
so we don't need to fixup x86_64 afterwards.
Reviewed-by: nrc
|
|\ \
| |/
| |
| |
| |
| | |
Conflicts:
src/gui/kernel/qcocoaview_mac_p.h
src/gui/widgets/qmainwindow.cpp
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Store whether qt was built with qt3support enabled in .qmake.cache. We
need this info to decide if QT3_SUPPORT should be defined when we
include qimage_p.h. Otherwise we get data corruption issues since the
size of struct in Qt is larger than the one we include in libqdirectfb.
Reviewed-by: Donald <qt-info@nokia.com>
|
| |
| |
| |
| |
| |
| |
| |
| | |
Reported downstream to Debian:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=525570
Reviewed-by: Alexis
Reviewed-by: Thiago
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
architechtures.
For example,
./configure -arch x86_64 -arch ppc requires D-Bus compiled for x86_64 and ppc
in order to enable it.
Task-number: 249881
Reviewed-by: nrc
|
|\ \
| |/
| |
| |
| | |
Conflicts:
src/gui/itemviews/qabstractitemview.cpp
|
| | |
|
|\ \
| |/
| |
| |
| | |
Conflicts:
src/gui/itemviews/qheaderview_p.h
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The nightly public snapshots contain bin/syncqt but do not
contain the include/ directory. Hence configure needs to
detect this case and run syncqt to create the include tree
so that configure actually passes
Signed-off-by: Thiago Macieira <thiago.macieira@nokia.com>
|
|\ \
| |/
| |
| |
| | |
Conflicts:
tests/auto/qpainterpath/tst_qpainterpath.cpp
|