I'm using Qt Creator in Ubuntu. It's installed from the repositories but as it is now, there is no way to step into the Qt sources when debugging.
How can I enable that?
Since Qt Creator uses gdb, you need to configure gdb. First thing to do is to install Qt debugging symbols:
apt-get install libqt4-dbg
Or, for Qt5:
apt-get install qtbase5-dbg # For the qtbase package
This will install the debugging symbols for Qt libraries. Older releases of Ubuntu had a silly bug that required additional trick to correct those symbol files, but in the current release it works fine.
This will make gdb step inside Qt methods, but it's no fun without sources. So we need sources which can be installed like this, assuming that the source repository is enabled in the APT:
apt-get source qt4-x11
ln -s qt4-x11-4.7.0 qt # a convenience symlink
Or, for Qt5:
apt-get source qtbase-opensource-src
# Make a link as above, if you wish
This will download the sources, unpack them into the current directory and patch them accordingly, no root privileges needed unless the current dir isn't writeable by the current user.
And the last thing is to inform gdb of the sources location, which is done by putting this in the ~/.gdbinit file:
dir ~/vita/qt/src/corelib
dir ~/vita/qt/src/gui
dir ~/vita/qt/src/network
dir ~/vita/qt/src/sql
Add modules and correct paths as needed. The convenience symlink is very useful here, so we don't have to edit this file each time we upgrade to a new Qt version. We only need to download the new sources, patch them and change the symlink.
Note that even we have installed the debugging symbols, we still use the release build of Qt libraries. This means that the code is highly optimized and will sometimes behave very strange when stepping inside Qt binaries. If it is a problem, then it is necessary to build Qt in debug mode, install it separately (say, in /usr/local/qt4-debug) and tell Qt Creator to use that particular installation.
The only way i made it work on Ubuntu is building Qt from sources with configure -debug.
Everything started to work like a charm afterwards.
Qt binary packages for Linux don't contain debug symbols and therefore the debugger doesn't know files or line numbers where to jump to. You need to build Qt yourself with -debug configure option if you want to be able to debug Qt code.
Instructions from Sergey Tachenov instructions would only work if you build your application against the Qt version that you can find from Ubuntu repositories.
Related
FIRST, it's not the issue that Qt CANNOT found target .so, qt can found it, but it failed to load it and produce not logs... on both terminal and gdb. It seems that qmake/make is using wrong library but I have only one qt installed in my cross compile environment that generated by make install.
I'm struggling for cross compiling qt5.10 for raspberry pi on Windows. I'm using msys2, gnutoolchains's raspberry chains.
The problem here is the examples built by the compiling process runs almost no problem, I only needs to add a qt.conf to fix the prefix override by msys2 and it can run and display something on my pi.
But when it comes to qt creator, or qmake out of cross compiling, it starts to show me:
This application failed to start because it could not find or load the Qt platform plugin "xcb"
Here are two problems, first, the platform plugin should not be xcb since I'm running without x, cross compiled version will use eglfs by default.
And, second, even I specific platform plugin to eglfs, it still tells me it cannot load eglfs.
I'm putting two version of programs in the same place.
qt5pi/examples/opengl/2dpainting $ ls
2dpainting glwidget.cpp helper.h widget.cpp window.h
2dpainting.pro glwidget.h main.cpp widget.h
2dpaint_my helper.cpp qt.conf window.cpp
2dpaint_my is compiled by qmake && make and 2dpainting is by qt's cross compiling process uses the same source.
I'm suspecting qt is adding something during it's cross compiling, but I'm not sure how it happened. Qt'wiki about raspberry pi contains nothing about this issue.
update
It looks even wired to me. I copied a running example from its folder to another and it also crash, things look like this..
pi#raspberrypi:/usr/local/qt5pi/examples/qt_test $ cp ../opengl/2dpainting/2dpainting .
pi#raspberrypi:/usr/local/qt5pi/examples/qt_test $ ./2dpainting
This application failed to start because it could not find or load the Qt platform plugin "xcb"
in "".
Available platform plugins are: eglfs, linuxfb, minimal, minimalegl, offscreen,vnc, webgl, xcb.
Reinstalling the application may fix this problem.
Aborted
pi#raspberrypi:/usr/local/qt5pi/examples/qt_test $ ../opengl/2dpainting/2dpainting
qt.qpa.egldeviceintegration: EGL device integration plugin keys: ("eglfs_brcm","eglfs_emu")
qt.qpa.egldeviceintegration: EGL device integration plugin keys (sorted): ("eglfs_brcm", "eglfs_emu")
qt.qpa.egldeviceintegration: Trying to load device EGL integration "eglfs_brcm"
qt.qpa.egldeviceintegration: Using EGL device integration "eglfs_brcm"
qt.qpa.input: Initializing tslib plugin "TsLib" ""
qt.qpa.input: tslib device is "/dev/input/event0"
I'm not sure, but it seems that the this step on qt'wiki introduced some qt5.7 files into lib folder.
sudo apt-get update
sudo apt-get build-dep qt4-x11
sudo apt-get build-dep libqt5gui5
sudo apt-get install libudev-dev libinput-dev libts-dev libxcb-xinerama0-dev libxcb-xinerama0
The problem is, qt's examples, when it on their own folders, can magically find (maybe qmake install did something?) and use the correct *.so, and so it works without problem.
But, when compiling with qmake, the program won't find the correct libraries so, newer libq*.so with older libQtXXXXX.so lead to this probleam. And that's how it uses a different qt version with compile.
And, that indicates the optional 00- in step 13 from the wiki:
[on RPi] Update the device to let the linker find the Qt libs:
echo /usr/local/qt5pi/lib | sudo tee /etc/ld.so.conf.d/qt5pi.conf
sudo ldconfig
If you're facing issues with running the example, try to use 00-qt5pi.conf instead of qt5pi.conf, to introduce proper order.
should be taken even the examples run without problem.
I have an application that I built using Qt Creator on Linux and want to deploy it now. However, I don't want to statically build it as I don't want it to be open-sourced. I tried the ldd ./YourExecutable command, however that only lists (and not add) the additional dependencies the application needs in order for it to run. My question is, how do I gather the necessary dependencies without having to individually look for these files? Is there a tool, such as windeployqt.exe on Windows, that I can use on Linux for the same purpose? Or is there a better approach than the one I'm thinking of?
Get Cygwin setup.exe: http://www.cygwin.com/
1.1. Run setup.exe and continue to package selection list.
1.2. Under Devel catagory select tools you need for compiling your source. For
example 'GNU make'.
1.3. Finish installing.
Get linux crosscompilers for cygwin:
"cygwin-gcc-linux.tar.bz2" (68.2 Mb).
md5sum: 340e91a346f5bb17e660db10e43005b8
These compilers are made with crosstool 0.28-rc37. This package contains:
gcc-3.3.4 and gcc-2.95.3 for i386 (glibc 2.1.3) and gcc-3.3.3 for amd64
(glibc 2.3.2).
Note! There is now newer version of GCC avaible with glibc 2.3.2:
"cygwin-gcc-3.3.6-glibc-2.3.2-linux.tar.bz2 (i386, x86_64)".
2.1. Copy 'cygwin-gcc-linux.tar.bz2' to 'c:\cygwin' or install directory which
you selected in setup.exe.
2.2. Open Cygwin shell and change directory to root with 'cd /'.
2.3. Uncompress to Cygwin root with command:
'tar -jxvf cygwin-gcc-linux.tar.bz2'.
Cross-compilers are installed under '/opt/crosstool'. You can use theim
directly or with commands: gcc-linux, g++-linux, gcc-linux-2.95,
g++-linux-2.95, gcc-linux-x86_64 and g++-linux-x86_64.
From: Cross-compiling on Windows for Linux
More info here.
It sounds like you want to use the shared library deployment option:
http://doc.qt.io/qt-5/linux-deployment.html#creating-the-application-package
Then if you wanted to go further than that, you could look into making a .rpm or a .deb .
There are lots of examples of qt projects that are now available on GitHub and have packages made. Usually for prebuilt binaries you need to make one for x86 and a separate one for x64.
Hope that helps.
Because I am not a Java enthusiast, I decided to use C++ and Qt for one of my projects. However, I came across the big cross-compiling Qt problem, and I am unable to produce an .exe file for Windows users.
My setup
Linux Ubuntu 12.04, with Wine and Qt. qmake -v gives the following output :
QMake version 2.01a
Using Qt version 4.8.1 in /usr/lib/x86_64-linux-gnu
I also have a MinGW32 compiler, which can be found at /usr/bin/i586-mingw32msvc-g++. My Wine drive_c folder contains the following Qt directories :
$HOME/.wine/drive_c/Qt/Qt5.2.0/5.2.0/Src
$HOME/.wine/drive_c/Qt/Qt5.2.0/5.2.0/mingw48_32
The mingw48_32 directory contains the necessary include/ and lib/ directories, which are used in my mkspec file, /usr/share/qt4/mkspecs/win32-x-g++/qmake.conf :
QMAKE_INCDIR_QT = /home/me/.wine/drive_c/Qt/Qt5.2.0/5.2.0/mingw48_32/include
QMAKE_LIBDIR_QT = /home/me/.wine/drive_c/Qt/Qt5.2.0/5.2.0/mingw48_32/lib
The problem
According to most guides I've found about Qt cross-compiling, my setup should be enough to run a simple :
qmake -spec win32-x-g++
make
wine /path/to/my/application.exe
But... nothing's linked. QApplication and every other symbol I use in my program are "not found". No QApplication, no QPushButton, no connect(), no SIGNAL(), no SLOT()...
My objective here is to successfully configure QtCreator to use this setup (in an independent build configuration), so that it can build a Linux executable (through the first and working configuration), and a Win32 .exe (through the MinGW setup above). However, I cannot modify a single build step in QtCreator :
Cannot add a "MinGW" toolchain : it is not available in the "Add" dropdown list.
Cannot change the -spec parameter value in the project build configurations panel. The field is non-editable.
Despite guides and solutions I found all over the Internet, my only solution so far is to send my source code to a virtual Windows machine, and have it create a new project with it. On this VM, I could probably compile for Windows... But of course, this doesn't actually sound like a real "solution" to me...
Is there any way Qt(Creator) has finally made cross-compiling easier now ? I'm getting a bit tired of "symbol not found" errors...
First,
sudo apt-get install mingw-w64
Then, check if Qt Creator finds the toolchain.
Next, until Ubuntu starts providing a mingw-w64-qt package, download the Qt source and build it. This is bound to get messy, and maybe even the simplest thing to do is to install WINE and use a Windows Qt version.
Trying to move a Qt 4.6.3 project from Windows to Debian Linux (Etch). I've installed Qt by downloading the latest version, then running configure, then make install as root. Yet when I run qmake_qt4 on my project's pri file, here's what I get:
uic: File generated with too old version of Qt Designer
Running uic -v returns 4.2.1. How come? Wasn't make install supposed to put the latest Qt on the system?
By default, Qt installs to /usr/local/Trolltech/Qt-4.6.3 and doesn't add it's bin folder to your PATH variable. Also, it's just named qmake.
Either call /usr/local/Trolltech/Qt-4.6.3/bin/qmake using the complete path or add the /usr/local/Trolltech/Qt-4.6.3/bin/ path to your PATH variable and ensure that it is listed there before any other path that may have a qmake. Best thing to do is to uninstall the the debian qt version if you don't need it elsewhere.
I installed the wxWidgets source code, compiled it and am linking the libraries thus obtained with my application code. Now I need to use OpenGL in my wxWidgets application. How do I enable this?
For building on Windows with project files:
Assume $(WXWIDGETSROOT) is the root directory of your wxWidgets installation.
Open the file $(WXWIDGETSROOT)\include\wx\msw\setup.h
Search for the #define for wxUSE_GLCANVAS.
Change its value from 0 to 1.
Recompile the library.
For building on Linux and other ./configure based platforms:
Just use ./configure --with-opengl
(A mashup answer from two partial answers given by others)
If you're using configure to build wxWidgets you just need to add --with-opengl to your command line.
Just to add a little bit... If you're on linux you need to watch the logs when running configure. If it can't find opengl dev packages then it will turn opengl off with one line of warning which is easy to miss.
run it like this to make it more obvious what development libraries you're actually missing (it looks like the --with-opengl is on by default in 3.0.0 and possibly earlier versions of wxwidgets, but it can't hurt to include it I suspect).
./configure --with-opengl > configure.log
Once configure can find all the dev libs you think you're going to use you need to rebuild wxwidgets:
make
sudo make install
I had to install these on linux mint to make wxwidget's configure happy as far as opengl was concerned (and should also work for ubuntu) to get the dev libs I needed.
sudo apt-get install mesa-common-dev
sudo apt-get install freeglut3-dev
(Assume $(WX_WIDGETS_ROOT) is the root directory of your wxWidgets installation.)
Open the file $(WX_WIDGETS_ROOT)\include\wx\msw\setup.h
Search and find the option wxUSE_GLCANVAS. Change its value from 0 to 1.
Recompile the library.