I am setting up a Qt project which will be cross-copiled for a given display (running on ARM/Linux). The display will come with a self-contained cross-build toolchain which comprises Qt libraries with a specific Qt version (probably 5.15).
Now I would like to use a specific "static" toolchain incl. Qt libraries to also generate x86-builds of this Qt project (for testing, simulation; building/running on Linux and maybe also on Windows).
I'd avoid using the GCC and Qt development files introduced by the Linux package manager (e.g. apt), because the are not tied to a specific Qt version, i.e. upgrading the OS would always come with the risk to introduce build problems or break compatibility completely.
Are there good-practice approaches to achieve this?
So far I see three options:
Use a docker image
Possible problems:
Finding a image with the correct Qt version
building the Qt application would then work, but to execute it, the compatible Qt LGPL shared libs need to be available to the Qt application binary
Use a x86 -> x86 "standalone" toolchain
Find some standalone x86-toolchain analog to those (x86 -> x86), but additionaly...
it would either have to come with the Qt development files in the correct version
or we would have to build the Qt sources with this toolchain, and then extend the toolchain with the Qt built artifacts
Would be the preferred solution (and currently seems to be the most feasible of the approaches I can think of).
Virtual machine
Possible problems:
Finding a OS version with the correct Qt version
Select a OS version that contains the correct Qt version via package manager.
Not very elegant.
Qt online installer
only provides a limited choice of Qt versions (as of today: 5.15.2, 6.2.4, 6.3.2, 6.4.0) which is probably subject to change in the future
Related
I know there are tons of informations out there on the web about my question.
I have searched and tried for at least 2 days to get my setup up and running. But there are many unclear points and i would be very happy if someone could help me with answering my questions.
My situation:
I have an i.MX6 Eval Board on my desk which is running uboot with Linux Kernel 4.9.2. There is no distro on top. Just the kernel with busybox.
Attached to the board is a LCD TFT which is accesed through the Framebuffer. There is no GPU in the Processor.
To compile uboot and the kernel, i use the linaro toolchain gcc-linaro-6.2.1-2016.11-x86_64_arm-linux-gnueabihf This toolchain is located at /opt/toolchain. The compilation of the kernel and uboot works without any problems.
Now i would like to start the application development with QT Creator. I have therefore read a lot of topics on various forums and sites on the web. Unfortunately i did not get it working.
First of all i have a confusion about the meaning of the title: http://doc.qt.io/qt-4.8/qt-embedded-install.html
There is a guide how to install "Installing Qt for Embedded Linux" Does this mean that this version is for an embedded linux, so it means that it has to be installed on the embedded device. Or does it mean that it guides you how to install QT to develop for a target system and install QT on an other system?
Before i wrote this question, i have already installed QT5.7 from their website. After some googling, i figured out that there are some changes in the windowsystem between 4.8 and 5.x. Since i dont have any GPU and therefore i dont have any OpenGL support, im not sure whether i should better go with 4.8 or whether i could use the 5.x version. Usually i tend to use the latest versions if possible.
Next i saw, that i have to define the sysroot directory. In the sysroot directory there have to be the compiled libs for qt for the arm architecture. Is there a precompiled version of these libs or do i have to compile them by my self? If yes, where to start?
EDIT:
In some tutorials they use the angstrom toolchain with a "qte" suffix My linaro toolchain does not have such an suffix. Is it also possible to use my linaro toolchain to build the qt applications?
Maybe you will notice that im new to QT and to the Embedded Linux world.
Anyhow i hope there are people out there who will be happy to share their knowledge about this topic with me.
Thanks.
I'm trying since days to write and develop programs on my host pc (Ubuntu 64bit) using qt-creator for my beaglebone and raspberry pi devices.
I looked for 'qt-creator cross compile beagle, raspberry etc...' and it leads me to hundred of links and tutorial on how to compile qt-creator for my target plattform. But seriously, that not my idea. All what I want is to simply use my pc for developing my programs and lately deploy them to the remote device.
So first of all a noob question: Is something wrong with my idea? Should I really compile the source of qt for beaglebone or raspberry on my pc for writing programs which lately should be run on that devices????
I will not compile qt-creator again for another plattform, so I don't know how to find the right way. A lot of informations on internet are misleading.
So after many hours spend look for the right information I come up with the following procedure:
first I installed a arm-gnu toolchain on my 64bit Ubuntu.
sudo apt-get install build-essential libc6-armhf-cross libc6-dev-armhf-cross binutils-arm-linux-gnueabihf linux-libc-dev-armhf-cross libstdc++6-armhf-cross gcc-4.8-arm-linux-gnueabihf g++-4.8-arm-linux-gnueabihf
then I installed qt-creator from the repository on my Ubuntu:
sudo apt-get install qtcreator
I tried to configure to use qt for compiling my projects using the toolchain for arm (see following pictures)
and here:
but it seems that qt-creator doesn't recognize the new kit when I start a new project:
Now: I m really confused, I could think that the problem is related to qt, but I m not sure, whether the way of thinking, that I can compile on ubuntu 64bit and then run my application on the remote device is right or not.
For this reason it would be nice if you can answer the question above.
UPDATE: What happen if I add a QT-Version in my Kit???
It happens that the gnueabi compiler conflicts with that options and the following problem arises:
I would like to say again that I don't need to use qt-library or something like that on my remote device. I need just to use the qt-IDE as IDE for my projects (mainly c or c++ plain projects)
You need a complete toolchain
A toolchain is just a name for a group of things:
Compiler for the target platform (or cross compiler in your case)
Sysroot: An image of the filesystem in the target platform
A cross compiling qmake
QtCreator and the kit
You are lacking a cross compiling qmake. Please see my answer at this other post.
Furthermore, you need a toolchain for the Pi, and another toolchain for the Beagle.
It looks like before the update in the question, you didn't even have a valid kit, so there was nothing the IDE could do. You need to create a kit, and set Qt Version to None - because whatever Qt version you have installed is not for the ARM target. So, as long as it'll be none, it should work then, assuming that you have installed a full ARM toolchain for the target, including the Linux-specific bits - since you're not building a bare-bones application, but an application that will run on Linux.
When creating a new project, make sure to select Non-Qt Project, either C or C++, and choose the CMake build system (not qmake or qbs).
To test this out, start with creating a kit for your host platform, without selecting a Qt version (set it to None), and make sure you can create a simple C project and build it (choose proper compilers etc). Once you have done that, you should have an idea how to proceed with building for your ARM Linux targets. It really shouldn't be problematic at all (it all "just works" in my experience), so you probably didn't install something major (related to the target toolchain, nothing to do with Qt nor Qt Creator).
In Derek Molloy's tutorial for the Beaglebone on this subject, he mentions at around minute 18 that installing QT directly through the command sudo apt-get install qtcreator, won't install all the features of the QT app.
Try downloading QT directly from Qt site.
I'm trying to learn how to use Qt for dev and have to say it's been a pain from the start.
I installed Qt Creator 5.4 version. Then followed this http://qt-project.org/wiki/Qt_for_beginners_Hello_World.
It defaulted to MSVS so I manually added MinGw. It has a red sign next to it saying that MinGW cannot compile code for the QT version.
I just want to use Qt5 and need an appropriate compiler for it. I'm used to using MinGW in Eclipse for C++ dev.
Your installation seems to be incomplete or inconsistent.
To work with Qt you actually need two things: One is a compiler toolchain (called the "kit" in Qt Creator), the other is the compiled Qt library, which is what is visible in settings under "Qt versions". Both need to fit together for Qt to work.
In case you want to use Qt with Visual Studio, you should select the precompiled Qt version corresponding to your (previously installed) VS version during install when you use the online installer. Or use the offline installer for your VS version. After install, everything finds together itself.
Or you want to use MinGW. Do yourself a favor and use the MinGW version which is installed by the Qt installer, which is different to the one from MinGW.org (it is the 32 bit version of the MinGW-w64 project, which is more up to date than the one from MinGW.org, and has some other subtleties). It is selectable in the installer under "tools". Additionally select the Qt version precompiled with MinGW. After install, select the MinGW Kit for your new project, and everything again will work from scratch. Again, you can use either the offline installer including MinGW-precompiled Qt or the online-installer which has all versions of precompiled selectable to perform that installation.
I need to port one of my old program to Windows. The current version uses Qt 3 on Linux.
I'd like to see it within few days only to do some tests.
Instead of porting my code on Qt 4 (no time now), is it possible to recompile on Windows with Qt 3? Do I also need minGW? Which problems may I have doing this porting?
I can't find Qt 3 for windows, can someone give me a link? or I can use the same qt source I use on Linux?
Yes, you can !
Download the qt-win-3.3.x-8 project on Sourceforge.
QT4 still has back-compatible QT3 classes, prefixed with Q3* (see http://qt-project.org/doc/qt-4.8/qt3support.html)
You should be able to compile & run you QT3 app on QT4 via those classes with minimal adaptations.
Qt-3.2 for Windows was also distributed in a CD bundled with the book "C++ GUI programming with Qt3" at no additional cost with a license for non-commercial use.
ISBN: 978-0131240728
The supported compilers were MSVC and Borland. MingW is not mentioned.
I am developing an application in Qt, but using D language (with QtD binding). I've noticed that my app crashes with Qt 4.7.x, so I need to use Qt 4.6.2 instead. However in my system Qt 4.7.2 is installed. Unfortunately I neither make QtD work with the latest Qt versions nor (I'm afraid) count on QtD developers...
The only thing I need to make my application install in system Qt 4.6.2 libs and use it, but let all the other applications still use Qt 4.7.2. Is it possible? If it is, how to do it then?
The answer to your question is the version of Qt is determined by the qmake you use to generate your Makefile.
/opt/QtSDK/Qt-4.6.2/bin/qmake
/opt/QtSDK/Qt-4.7.3/bin/qmake
Each will use the library in the directory.
It is possible to install several versions of the library into one system, and all package managers (rpm, deb) support this out of the box.
These libraries will be just differently named. For example, if id do
ls /usr/lib/ | grep libcurl
on my system, I'll get:
libcurl-gnutls.so.3
libcurl-gnutls.so.4
libcurl-gnutls.so.4.2.0
libcurl.so.3
libcurl.so.4
libcurl.so.4.2.0
, so, different version of the same library happily live together.
All you need to do is to link against the desired version of QT library. You need to sepcify it in linker options (don't remember the exact option).