/**************************************************************************** ** ** Copyright (C) 2009 Nokia Corporation and/or its subsidiary(-ies). ** Contact: Qt Software Information (qt-info@nokia.com) ** ** This file is part of the documentation of the Qt Toolkit. ** ** $QT_BEGIN_LICENSE:LGPL$ ** Commercial Usage ** Licensees holding valid Qt Commercial licenses may use this file in ** accordance with the Qt Commercial License Agreement provided with the ** Software or, alternatively, in accordance with the terms contained in ** a written agreement between you and Nokia. ** ** GNU Lesser General Public License Usage ** Alternatively, this file may be used under the terms of the GNU Lesser ** General Public License version 2.1 as published by the Free Software ** Foundation and appearing in the file LICENSE.LGPL included in the ** packaging of this file. Please review the following information to ** ensure the GNU Lesser General Public License version 2.1 requirements ** will be met: http://www.gnu.org/licenses/old-licenses/lgpl-2.1.html. ** ** In addition, as a special exception, Nokia gives you certain ** additional rights. These rights are described in the Nokia Qt LGPL ** Exception version 1.0, included in the file LGPL_EXCEPTION.txt in this ** package. ** ** GNU General Public License Usage ** Alternatively, this file may be used under the terms of the GNU ** General Public License version 3.0 as published by the Free Software ** Foundation and appearing in the file LICENSE.GPL included in the ** packaging of this file. Please review the following information to ** ensure the GNU General Public License version 3.0 requirements will be ** met: http://www.gnu.org/copyleft/gpl.html. ** ** If you are unsure which license is appropriate for your use, please ** contact the sales department at qt-sales@nokia.com. ** $QT_END_LICENSE$ ** ****************************************************************************/ /**************************************************************************** ** ** Copyright (C) 2009 Nokia Corporation and/or its subsidiary(-ies). ** Contact: Qt Software Information (qt-info@nokia.com) ** ** This file is part of the Qt GUI Toolkit. ** EDITIONS: FREE, PROFESSIONAL, ENTERPRISE ** ****************************************************************************/ /*! \group deployment \title Deploying Qt Applications \ingroup buildsystem Deploying an Qt application does not require any C++ programming. All you need to do is to build Qt and your application in release mode, following the procedures described in this documentation. We will demonstrate the procedures in terms of deploying the \l {tools/plugandpaint}{Plug & Paint} application that is provided in Qt's examples directory. \section1 Static vs. Shared Libraries There are two ways of deploying an application: \list \o Static Linking \o Shared Libraries (Frameworks on Mac) \endlist Static linking results in a stand-alone executable. The advantage is that you will only have a few files to deploy. The disadvantages are that the executables are large and with no flexibility (i.e a new version of the application, or of Qt, will require that the deployment process is repeated), and that you cannot deploy plugins. To deploy plugin-based applications, you can use the shared library approach. Shared libraries also provide smaller, more flexible executables. For example, using the shared library approach, the user is able to independently upgrade the Qt library used by the application. Another reason why you might want to use the shared library approach, is if you want to use the same Qt libraries for a family of applications. In fact, if you download the binary installation of Qt, you get Qt as a shared library. The disadvantage with the shared library approach is that you will get more files to deploy. For more information, see \l{sharedlibrary.html}{Creating Shared Libraries}. \section1 Deploying Qt's Libraries \table \header \o {4,1} Qt's Libraries \row \o \l {QtAssistant} \o \l {QAxContainer} \o \l {QAxServer} \o \l {QtCore} \row \o \l {QtDBus} \o \l {QtDesigner} \o \l {QtGui} \o \l {QtHelp} \row \o \l {QtNetwork} \o \l {QtOpenGL} \o \l {QtScript} \o \l {QtScriptTools} \row \o \l {QtSql} \o \l {QtSvg} \o \l {QtWebKit} \o \l {QtXml} \row \o \l {QtXmlPatterns} \o \l {Phonon Module}{Phonon} \o \l {Qt3Support} \endtable Since Qt is not a system library, it has to be redistributed along with your application; the minimum is to redistribute the run-time of the libraries used by the application. Using static linking, however, the Qt run-time is compiled into the executable. In particular, you will need to deploy Qt plugins, such as JPEG support or SQL drivers. For more information about plugins, see the \l {plugins-howto.html}{How to Create Qt Plugins} documentation. When deploying an application using the shared library approach you must ensure that the Qt libraries will use the correct path to find the Qt plugins, documentation, translation etc. To do this you can use a \c qt.conf file. For more information, see the \l {Using qt.conf} documentation. Depending on configuration, compiler specific libraries must be redistributed as well. For more information, see the platform specific Application Dependencies sections: \l {deployment-x11.html#application-dependencies}{X11}, \l {deployment-windows.html#application-dependencies}{Windows}, \l {deployment-mac.html#application-dependencies}{Mac}. \section1 Licensing Some of Qt's libraries are based on third party libraries that are not licensed using the same dual-license model as Qt. As a result, care must be taken when deploying applications that use these libraries, particularly when the application is statically linked to them. The following table contains an inexhaustive summary of the issues you should be aware of. \table \header \o Qt Library \o Dependency \o Licensing Issue \row \o QtHelp \o CLucene \o The version of clucene distributed with Qt is licensed under the GNU LGPL version 2.1 or later. This has implications for developers of closed source applications. Please see \l{QtHelp Module#License Information}{the QtHelp module documentation} for more information. \row \o QtNetwork \o OpenSSL \o Some configurations of QtNetwork use OpenSSL at run-time. Deployment of OpenSSL libraries is subject to both licensing and export restrictions. More information can be found in the \l{Secure Sockets Layer (SSL) Classes} documentation. \row \o QtWebKit \o WebKit \o WebKit is licensed under the GNU LGPL version 2 or later. This has implications for developers of closed source applications. Please see \l{QtWebKit Module#License Information}{the QtWebKit module documentation} for more information. \row \o \l{Phonon Module}{Phonon} \o Phonon \o Phonon relies on the native multimedia engines on different platforms. Phonon itself is licensed under the GNU LGPL version 2. Please see \l{Phonon Module#License Information}{the Phonon module documentation} for more information on licensing and the \l{Phonon Overview#Backends}{Phonon Overview} for details of the backends in use on different platforms. \endtable \section1 Platform-Specific Notes The procedure of deploying Qt applications is different for the various platforms: \list \o \l{Deploying an Application on X11 Platforms}{Qt for X11 Platforms} \o \l{Deploying an Application on Windows}{Qt for Windows} \o \l{Deploying an Application on Mac OS X}{Qt for Mac OS X} \o \l{Deploying Qt for Embedded Linux Applications}{Qt for Embedded Linux} \endlist \sa Installation {Window System Specific Notes} */ /*! \page deployment-x11.html \contentspage Deploying Qt Applications \title Deploying an Application on X11 Platforms \ingroup deployment Due to the proliferation of Unix systems (commercial Unices, Linux distributions, etc.), deployment on Unix is a complex topic. Before we start, be aware that programs compiled for one Unix flavor will probably not run on a different Unix system. For example, unless you use a cross-compiler, you cannot compile your application on Irix and distribute it on AIX. Contents: \tableofcontents This documentation will describe how to determine which files you should include in your distribution, and how to make sure that the application will find them at run-time. We will demonstrate the procedures in terms of deploying the \l {tools/plugandpaint}{Plug & Paint} application that is provided in Qt's examples directory. \section1 Static Linking Static linking is often the safest and easiest way to distribute an application on Unix since it relieves you from the task of distributing the Qt libraries and ensuring that they are located in the default search path for libraries on the target system. \section2 Building Qt Statically To use this approach, you must start by installing a static version of the Qt library: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 0 We specify the prefix so that we do not overwrite the existing Qt installation. The example above only builds the Qt libraries, i.e. the examples and Qt Designer will not be built. When \c make is done, you will find the Qt libraries in the \c /path/to/Qt/lib directory. When linking your application against static Qt libraries, note that you might need to add more libraries to the \c LIBS line in your project file. For more information, see the \l {Application Dependencies} section. \section2 Linking the Application to the Static Version of Qt Once Qt is built statically, the next step is to regenerate the makefile and rebuild the application. First, we must go into the directory that contains the application: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 1 Now run qmake to create a new makefile for the application, and do a clean build to create the statically linked executable: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 2 You probably want to link against the release libraries, and you can specify this when invoking \c qmake. Note that we must set the path to the static Qt that we just built. To check that the application really links statically with Qt, run the \c ldd tool (available on most Unices): \snippet doc/src/snippets/code/doc_src_deployment.qdoc 3 Verify that the Qt libraries are not mentioned in the output. Now, provided that everything compiled and linked without any errors, we should have a \c plugandpaint file that is ready for deployment. One easy way to check that the application really can be run stand-alone is to copy it to a machine that doesn't have Qt or any Qt applications installed, and run it on that machine. Remember that if your application depends on compiler specific libraries, these must still be redistributed along with your application. For more information, see the \l {Application Dependencies} section. The \l {tools/plugandpaint}{Plug & Paint} example consists of several components: The core application (\l {tools/plugandpaint}{Plug & Paint}), and the \l {tools/plugandpaintplugins/basictools}{Basic Tools} and \l {tools/plugandpaintplugins/extrafilters}{Extra Filters} plugins. Since we cannot deploy plugins using the static linking approach, the executable we have prepared so far is incomplete. The application will run, but the functionality will be disabled due to the missing plugins. To deploy plugin-based applications we should use the shared library approach. \section1 Shared Libraries We have two challenges when deploying the \l {tools/plugandpaint}{Plug & Paint} application using the shared libraries approach: The Qt runtime has to be correctly redistributed along with the application executable, and the plugins have to be installed in the correct location on the target system so that the application can find them. \section2 Building Qt as a Shared Library We assume that you already have installed Qt as a shared library, which is the default when installing Qt, in the \c /path/to/Qt directory. For more information on how to build Qt, see the \l {Installation} documentation. \section2 Linking the Application to Qt as a Shared Library After ensuring that Qt is built as a shared library, we can build the \l {tools/plugandpaint}{Plug & Paint} application. First, we must go into the directory that contains the application: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 4 Now run qmake to create a new makefile for the application, and do a clean build to create the dynamically linked executable: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 5 This builds the core application, the following will build the plugins: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 6 If everything compiled and linked without any errors, we will get a \c plugandpaint executable and the \c libpnp_basictools.so and \c libpnp_extrafilters.so plugin files. \section2 Creating the Application Package There is no standard package management on Unix, so the method we present below is a generic solution. See the documentation for your target system for information on how to create a package. To deploy the application, we must make sure that we copy the relevant Qt libraries (corresponding to the Qt modules used in the application) as well as the executable to the same directory. Remember that if your application depends on compiler specific libraries, these must also be redistributed along with your application. For more information, see the \l {Application Dependencies} section. We'll cover the plugins shortly, but the main issue with shared libraries is that you must ensure that the dynamic linker will find the Qt libraries. Unless told otherwise, the dynamic linker doesn't search the directory where your application resides. There are many ways to solve this: \list \o You can install the Qt libraries in one of the system library paths (e.g. \c /usr/lib on most systems). \o You can pass a predetermined path to the \c -rpath command-line option when linking the application. This will tell the dynamic linker to look in this directory when starting your application. \o You can write a startup script for your application, where you modify the dynamic linker configuration (e.g. adding your application's directory to the \c LD_LIBRARY_PATH environment variable. \note If your application will be running with "Set user ID on execution," and if it will be owned by root, then LD_LIBRARY_PATH will be ignored on some platforms. In this case, use of the LD_LIBRARY_PATH approach is not an option). \endlist The disadvantage of the first approach is that the user must have super user privileges. The disadvantage of the second approach is that the user may not have privileges to install into the predetemined path. In either case, the users don't have the option of installing to their home directory. We recommend using the third approach since it is the most flexible. For example, a \c plugandpaint.sh script will look like this: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 7 By running this script instead of the executable, you are sure that the Qt libraries will be found by the dynamic linker. Note that you only have to rename the script to use it with other applications. When looking for plugins, the application searches in a plugins subdirectory inside the directory of the application executable. Either you have to manually copy the plugins into the \c plugins directory, or you can set the \c DESTDIR in the plugins' project files: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 8 An archive distributing all the Qt libraries, and all the plugins, required to run the \l {tools/plugandpaint}{Plug & Paint} application, would have to include the following files: \table 100% \header \o Component \o {2, 1} File Name \row \o The executable \o {2, 1} \c plugandpaint \row \o The script to run the executable \o {2, 1} \c plugandpaint.sh \row \o The Basic Tools plugin \o {2, 1} \c plugins\libpnp_basictools.so \row \o The ExtraFilters plugin \o {2, 1} \c plugins\libpnp_extrafilters.so \row \o The Qt Core module \o {2, 1} \c libQtCore.so.4 \row \o The Qt GUI module \o {2, 1} \c libQtGui.so.4 \endtable On most systems, the extension for shared libraries is \c .so. A notable exception is HP-UX, which uses \c .sl. Remember that if your application depends on compiler specific libraries, these must still be redistributed along with your application. For more information, see the \l {Application Dependencies} section. To verify that the application now can be successfully deployed, you can extract this archive on a machine without Qt and without any compiler installed, and try to run it, i.e. run the \c plugandpaint.sh script. An alternative to putting the plugins in the \c plugins subdirectory is to add a custom search path when you start your application using QApplication::addLibraryPath() or QApplication::setLibraryPaths(). \snippet doc/src/snippets/code/doc_src_deployment.qdoc 9 \section1 Application Dependencies \section2 Additional Libraries To find out which libraries your application depends on, run the \c ldd tool (available on most Unices): \snippet doc/src/snippets/code/doc_src_deployment.qdoc 10 This will list all the shared library dependencies for your application. Depending on configuration, these libraries must be redistributed along with your application. In particular, the standard C++ library must be redistributed if you're compiling your application with a compiler that is binary incompatible with the system compiler. When possible, the safest solution is to link against these libraries statically. You will probably want to link dynamically with the regular X11 libraries, since some implementations will try to open other shared libraries with \c dlopen(), and if this fails, the X11 library might cause your application to crash. It's also worth mentioning that Qt will look for certain X11 extensions, such as Xinerama and Xrandr, and possibly pull them in, including all the libraries that they link against. If you can't guarantee the presence of a certain extension, the safest approach is to disable it when configuring Qt (e.g. \c {./configure -no-xrandr}). FontConfig and FreeType are other examples of libraries that aren't always available or that aren't always binary compatible. As strange as it may sound, some software vendors have had success by compiling their software on very old machines and have been very careful not to upgrade any of the software running on them. When linking your application against the static Qt libraries, you must explicitly link with the dependent libraries mentioned above. Do this by adding them to the \c LIBS variable in your project file. \section2 Qt Plugins Your application may also depend on one or more Qt plugins, such as the JPEG image format plugin or a SQL driver plugin. Be sure to distribute any Qt plugins that you need with your application, and note that each type of plugin should be located within a specific subdirectory (such as \c imageformats or \c sqldrivers) within your distribution directory, as described below. \note If you are deploying an application that uses QtWebKit to display HTML pages from the World Wide Web, you should include all text codec plugins to support as many HTML encodings possible. The search path for Qt plugins (as well as a few other paths) is hard-coded into the QtCore library. By default, the first plugin search path will be hard-coded as \c /path/to/Qt/plugins. As mentioned above, using pre-determined paths has certain disadvantages, so you need to examine various alternatives to make sure that the Qt plugins are found: \list \o \l{qt-conf.html}{Using \c qt.conf}. This is the recommended approach since it provides the most flexibility. \o Using QApplication::addLibraryPath() or QApplication::setLibraryPaths(). \o Using a third party installation utility or the target system's package manager to change the hard-coded paths in the QtCore library. \endlist The \l{How to Create Qt Plugins} document outlines the issues you need to pay attention to when building and deploying plugins for Qt applications. */ /*! \page deployment-windows.html \contentspage Deploying Qt Applications \title Deploying an Application on Windows \ingroup deployment This documentation will describe how to determine which files you should include in your distribution, and how to make sure that the application will find them at run-time. We will demonstrate the procedures in terms of deploying the \l {tools/plugandpaint}{Plug & Paint} application that is provided in Qt's examples directory. Contents: \tableofcontents \section1 Static Linking If you want to keep things simple by only having a few files to deploy, i.e. a stand-alone executable with the associated compiler specific DLLs, then you must build everything statically. \section2 Building Qt Statically Before we can build our application we must make sure that Qt is built statically. To do this, go to a command prompt and type the following: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 11 Remember to specify any other options you need, such as data base drivers, as arguments to \c configure. Once \c configure has finished, type the following: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 12 This will build Qt statically. Note that unlike with a dynamic build, building Qt statically will result in libraries without version numbers; e.g. \c QtCore4.lib will be \c QtCore.lib. Also, we have used \c nmake in all the examples, but if you use MinGW you must use \c mingw32-make instead. \note If you later need to reconfigure and rebuild Qt from the same location, ensure that all traces of the previous configuration are removed by entering the build directory and typing \c{nmake distclean} before running \c configure again. \section2 Linking the Application to the Static Version of Qt Once Qt has finished building we can build the \l {tools/plugandpaint}{Plug & Paint} application. First we must go into the directory that contains the application: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 13 We must then run \c qmake to create a new makefile for the application, and do a clean build to create the statically linked executable: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 14 You probably want to link against the release libraries, and you can specify this when invoking \c qmake. Now, provided that everything compiled and linked without any errors, we should have a \c plugandpaint.exe file that is ready for deployment. One easy way to check that the application really can be run stand-alone is to copy it to a machine that doesn't have Qt or any Qt applications installed, and run it on that machine. Remember that if your application depends on compiler specific libraries, these must still be redistributed along with your application. You can check which libraries your application is linking against by using the \c depends tool. For more information, see the \l {Application Dependencies} section. The \l {tools/plugandpaint}{Plug & Paint} example consists of several components: The application itself (\l {tools/plugandpaint}{Plug & Paint}), and the \l {tools/plugandpaintplugins/basictools}{Basic Tools} and \l {tools/plugandpaintplugins/extrafilters}{Extra Filters} plugins. Since we cannot deploy plugins using the static linking approach, the application we have prepared is incomplete. It will run, but the functionality will be disabled due to the missing plugins. To deploy plugin-based applications we should use the shared library approach. \section1 Shared Libraries We have two challenges when deploying the \l {tools/plugandpaint}{Plug & Paint} application using the shared libraries approach: The Qt runtime has to be correctly redistributed along with the application executable, and the plugins have to be installed in the correct location on the target system so that the application can find them. \section2 Building Qt as a Shared Library We assume that you already have installed Qt as a shared library, which is the default when installing Qt, in the \c C:\path\to\Qt directory. For more information on how to build Qt, see the \l {Installation} documentation. \section2 Linking the Application to Qt as a Shared Library After ensuring that Qt is built as a shared library, we can build the \l {tools/plugandpaint}{Plug & Paint} application. First, we must go into the directory that contains the application: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 15 Now run \c qmake to create a new makefile for the application, and do a clean build to create the dynamically linked executable: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 16 This builds the core application, the following will build the plugins: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 17 If everything compiled and linked without any errors, we will get a \c plugandpaint.exe executable and the \c pnp_basictools.dll and \c pnp_extrafilters.dll plugin files. \section2 Creating the Application Package To deploy the application, we must make sure that we copy the relevant Qt DLL (corresponding to the Qt modules used in the application) as well as the executable to the same directory in the \c release subdirectory. Remember that if your application depends on compiler specific libraries, these must be redistributed along with your application. You can check which libraries your application is linking against by using the \c depends tool. For more information, see the \l {Application Dependencies} section. We'll cover the plugins shortly, but first we'll check that the application will work in a deployed environment: Either copy the executable and the Qt DLLs to a machine that doesn't have Qt or any Qt applications installed, or if you want to test on the build machine, ensure that the machine doesn't have Qt in its environment. If the application starts without any problems, then we have successfully made a dynamically linked version of the \l {tools/plugandpaint}{Plug & Paint} application. But the application's functionality will still be missing since we have not yet deployed the associated plugins. Plugins work differently to normal DLLs, so we can't just copy them into the same directory as our application's executable as we did with the Qt DLLs. When looking for plugins, the application searches in a \c plugins subdirectory inside the directory of the application executable. So to make the plugins available to our application, we have to create the \c plugins subdirectory and copy over the relevant DLLs: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 18 An archive distributing all the Qt DLLs and application specific plugins required to run the \l {tools/plugandpaint}{Plug & Paint} application, would have to include the following files: \table 100% \header \o Component \o {2, 1} File Name \row \o The executable \o {2, 1} \c plugandpaint.exe \row \o The Basic Tools plugin \o {2, 1} \c plugins\pnp_basictools.dll \row \o The ExtraFilters plugin \o {2, 1} \c plugins\pnp_extrafilters.dll \row \o The Qt Core module \o {2, 1} \c qtcore4.dll \row \o The Qt GUI module \o {2, 1} \c qtgui4.dll \endtable In addition, the archive must contain the following compiler specific libraries depending on your version of Visual Studio: \table 100% \header \o \o VC++ 6.0 \o VC++ 7.1 (2003) \o VC++ 8.0 (2005) \row \o The C run-time \o \c msvcrt.dll \o \c msvcr71.dll \o \c msvcr80.dll \row \o The C++ run-time \o \c msvcp60.dll \o \c msvcp71.dll \o \c msvcp80.dll \endtable To verify that the application now can be successfully deployed, you can extract this archive on a machine without Qt and without any compiler installed, and try to run it. An alternative to putting the plugins in the plugins subdirectory is to add a custom search path when you start your application using QApplication::addLibraryPath() or QApplication::setLibraryPaths(). \snippet doc/src/snippets/code/doc_src_deployment.qdoc 19 One benefit of using plugins is that they can easily be made available to a whole family of applications. It's often most convenient to add the path in the application's \c main() function, right after the QApplication object is created. Once the path is added, the application will search it for plugins, in addition to looking in the \c plugins subdirectory in the application's own directory. Any number of additional paths can be added. \section2 Visual Studio 2005 Onwards When deploying an application compiled with Visual Studio 2005 onwards, there are some additional steps to be taken. First, we need to copy the manifest file created when linking the application. This manifest file contains information about the application's dependencies on side-by-side assemblies, such as the runtime libraries. The manifest file needs to be copied into the \bold same folder as the application executable. You do not need to copy the manifest files for shared libraries (DLLs), since they are not used. If the shared library has dependencies that are different from the application using it, the manifest file needs to be embedded into the DLL binary. Since Qt 4.1.3, the follwoing \c CONFIG options are available for embedding manifests: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 20 To use the options, add \snippet doc/src/snippets/code/doc_src_deployment.qdoc 21 to your .pro file. The \c embed_manifest_dll option is enabled by default. You can find more information about manifest files and side-by-side assemblies at the \l {http://msdn.microsoft.com/en-us/library/aa376307.aspx}{MSDN website}. There are two ways to include the run time libraries: by bundling them directly with your application or by installing them on the end-user's system. To bundle the run time libraries with your application, copy the directory \snippet doc/src/snippets/code/doc_src_deployment.qdoc 22 into the folder where your executable is, so that you are including a \c Microsoft.VC80.CRT directory alongside your application's executable. If you are bundling the runtimes and need to deploy plugins as well, you have to remove the manifest from the plugins (embedded as a resource) by adding the following line to the \c{.pro} file of the plugins you are compiling: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 23 \warning If you skip the step above, the plugins will not load on some systems. To install the runtime libraries on the end-user's system, you need to include the appropriate Visual C++ Redistributable Package (VCRedist) executable with your application and ensure that it is executed when the user installs your application. For example, on an 32-bit x86-based system, you would include the \l{http://www.microsoft.com/downloads/details.aspx?FamilyId=32BC1BEE-A3F9-4C13-9C99-220B62A191EE}{vcredist_x86.exe} executable. The \l{http://www.microsoft.com/downloads/details.aspx?familyid=526BF4A7-44E6-4A91-B328-A4594ADB70E5}{vcredist_IA64.exe} and \l{http://www.microsoft.com/downloads/details.aspx?familyid=90548130-4468-4BBC-9673-D6ACABD5D13B}{vcredist_x64.exe} executables provide the appropriate libraries for the IA64 and 64-bit x86 architectures, respectively. \note The application you ship must be compiled with exactly the same compiler version against the same C runtime version. This prevents deploying errors caused by different versions of the C runtime libraries. \section1 Application Dependencies \section2 Additional Libraries Depending on configuration, compiler specific libraries must be redistributed along with your application. You can check which libraries your application is linking against by using the \l{Dependency Walker} tool. All you need to do is to run it like this: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 24 This will provide a list of the libraries that your application depends on and other information. \image deployment-windows-depends.png When looking at the release build of the Plug & Paint executable (\c plugandpaint.exe) with the \c depends tool, the tool lists the following immediate dependencies to non-system libraries: \table 100% \header \o Qt \o VC++ 6.0 \o VC++ 7.1 (2003) \o VC++ 8.0 (2005) \o MinGW \row \o \list \o QTCORE4.DLL - The QtCore runtime \o QTGUI4.DLL - The QtGui runtime \endlist \o \list \o MSVCRT.DLL - The C runtime \o MSVCP60.DLL - The C++ runtime (only when STL is installed) \endlist \o \list \o MSVCR71.DLL - The C runtime \o MSVCP71.DLL - The C++ runtime (only when STL is installed) \endlist \o \list \o MSVCR80.DLL - The C runtime \o MSVCP80.DLL - The C++ runtime (only when STL is installed) \endlist \o \list \o MINGWM10.DLL - The MinGW run-time \endlist \endtable When looking at the plugin DLLs the exact same dependencies are listed. \section2 Qt Plugins Your application may also depend on one or more Qt plugins, such as the JPEG image format plugin or a SQL driver plugin. Be sure to distribute any Qt plugins that you need with your application, and note that each type of plugin should be located within a specific subdirectory (such as \c imageformats or \c sqldrivers) within your distribution directory, as described below. \note If you are deploying an application that uses QtWebKit to display HTML pages from the World Wide Web, you should include all text codec plugins to support as many HTML encodings possible. The search path for Qt plugins is hard-coded into the QtCore library. By default, the plugins subdirectory of the Qt installation is the first plugin search path. However, pre-determined paths like the default one have certain disadvantages. For example, they may not exist on the target machine. For that reason, you need to examine various alternatives to make sure that the Qt plugins are found: \list \o \l{qt-conf.html}{Using \c qt.conf}. This approach is the recommended if you have executables in different places sharing the same plugins. \o Using QApplication::addLibraryPath() or QApplication::setLibraryPaths(). This approach is recommended if you only have one executable that will use the plugin. \o Using a third party installation utility to change the hard-coded paths in the QtCore library. \endlist If you add a custom path using QApplication::addLibraryPath it could look like this: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 54 Then qApp->libraryPaths() would return something like this: "C:/customPath/plugins " "C:/Qt/4.5.1/plugins" "E:/myApplication/directory/" The executable will look for the plugins in these directories and the same order as the QStringList returned by qApp->libraryPaths(). The newly added path is prepended to the qApp->libraryPaths() which means that it will be searched through first. However, if you use qApp->setLibraryPaths(), you will be able to determend which paths and in which order they will be searched. The \l{How to Create Qt Plugins} document outlines the issues you need to pay attention to when building and deploying plugins for Qt applications. */ /*! \page deployment-mac.html \contentspage Deploying Qt Applications \title Deploying an Application on Mac OS X \ingroup deployment Starting with version 4.5, Qt now includes a \l {macdeploy}{deployment tool} that automates the prodecures described in this document. This documentation will describe how to create a bundle, and how to make sure that the application will find the resources it needs at run-time. We will demonstrate the procedures in terms of deploying the \l {tools/plugandpaint}{Plug & Paint} application that is provided in Qt's examples directory. \tableofcontents \section1 The Bundle On the Mac, a GUI application must be built and run from a bundle. A bundle is a directory structure that appears as a single entity when viewed in the Finder. A bundle for an application typcially contains the executable and all the resources it needs. See the image below: \image deployment-mac-bundlestructure.png The bundle provides many advantages to the user. One primary advantage is that, since it is a single entity, it allows for drag-and-drop installation. As a programmer you can access bundle information in your own code. This is specific to Mac OS X and beyond the scope of this document. More information about bundles is available on \l {http://developer.apple.com/documentation/CoreFoundation/Conceptual/CFBundles/index.html}{Apple's Developer Website}. A Qt command line application on Mac OS X works similar to a command line application on Unix and Windows. You probably don't want to run it in a bundle: Add this to your application's .pro: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 26 This will tell \c qmake not to put the executable inside a bundle. Please refer to the \l{Deploying an Application on X11 Platforms}{X11 deployment documentation} for information about how to deploy these "bundle-less" applications. \section1 Xcode We will only concern ourselves with command-line tools here. While it is possible to use Xcode for this, Xcode has changed enough between each version that it makes it difficult to document it perfectly for each version. A future version of this document may include more information for using Xcode in the deployment process. \section1 Static Linking If you want to keep things simple by only having a few files to deploy, then you must build everything statically. \section2 Building Qt Statically Start by installing a static version of the Qt library. Remember that you will not be able to use plugins and you must build in all the image formats, SQL drivers, etc.. \snippet doc/src/snippets/code/doc_src_deployment.qdoc 27 You can check the various options that are available by running \c configure -help. \section2 Linking the Application to the Static Version of Qt Once Qt is built statically, the next step is to regenerate the makefile and rebuild the application. First, we must go into the directory that contains the application: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 28 Now run \c qmake to create a new makefile for the application, and do a clean build to create the statically linked executable: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 29 You probably want to link against the release libraries, and you can specify this when invoking \c qmake. If you have Xcode Tools 1.5 or higher installed, you may want to take advantage of "dead code stripping" to reduce the size of your binary even more. You can do this by passing \c {LIBS+= -dead_strip} to \c qmake in addition to the \c {-config release} parameter. This doesn't have as large an effect if you are using GCC 4, since Qt will then have function visibility hints built-in, but if you use GCC 3.3, it could make a difference. Now, provided that everything compiled and linked without any errors, we should have a \c plugandpaint.app bundle that is ready for deployment. One easy way to check that the application really can be run stand-alone is to copy the bundle to a machine that doesn't have Qt or any Qt applications installed, and run the application on that machine. You can check what other libraries your application links to using the \c otool: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 30 Here is what the output looks like for the static \l {tools/plugandpaint}{Plug & Paint}: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 31 For more information, see the \l {Application Dependencies} section. If you see \e Qt libraries in the output, it probably means that you have both dynamic and static Qt libraries installed on your machine. The linker will always choose dynamic over static. There are two solutions: Either move your Qt dynamic libraries (\c .dylibs) away to another directory while you link the application and then move them back, or edit the \c Makefile and replace link lines for the Qt libraries with the absolute path to the static libraries. For example, replace \snippet doc/src/snippets/code/doc_src_deployment.qdoc 32 with \snippet doc/src/snippets/code/doc_src_deployment.qdoc 33 The \l {tools/plugandpaint}{Plug & Paint} example consists of several components: The core application (\l {tools/plugandpaint}{Plug & Paint}), and the \l {tools/plugandpaintplugins/basictools}{Basic Tools} and \l {tools/plugandpaintplugins/extrafilters}{Extra Filters} plugins. Since we cannot deploy plugins using the static linking approach, the bundle we have prepared so far is incomplete. The application will run, but the functionality will be disabled due to the missing plugins. To deploy plugin-based applications we should use the framework approach. \section1 Frameworks We have two challenges when deploying the \l {tools/plugandpaint}{Plug & Paint} application using frameworks: The Qt runtime has to be correctly redistributed along with the application bundle, and the plugins have to be installed in the correct location so that the application can find them. When distributing Qt with your application using frameworks, you have two options: You can either distribute Qt as a private framework within your application bundle, or you can distribute Qt as a standard framework (alternatively use the Qt frameworks in the installed binary). These two approaches are essentially the same. The latter option is good if you have many Qt applications and you would prefer to save memory. The former is good if you have Qt built in a special way, or want to make sure the framework is there. It just comes down to where you place the Qt frameworks. \section2 Building Qt as Frameworks We assume that you already have installed Qt as frameworks, which is the default when installing Qt, in the /path/to/Qt directory. For more information on how to build Qt, see the \l Installation documentation. When installing, the identification name of the frameworks will also be set. The identification name is what the dynamic linker (\c dyld) uses to find the libraries for your application. \section2 Linking the Application to Qt as Frameworks After ensuring that Qt is built as frameworks, we can build the \l {tools/plugandpaint}{Plug & Paint} application. First, we must go into the directory that contains the application: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 34 Now run qmake to create a new makefile for the application, and do a clean build to create the dynamically linked executable: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 35 This builds the core application, the following will build the plugins: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 36 Now run the \c otool for the Qt frameworks, for example Qt Gui: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 37 You will get the following output: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 38 For the Qt frameworks, the first line (i.e. \c {path/to/Qt/lib/QtGui.framework/Versions/4/QtGui (compatibility version 4.0.0, current version 4.0.1)}) becomes the framework's identification name which is used by the dynamic linker (\c dyld). But when you are deploying the application, your users may not have the Qt frameworks installed in the specified location. For that reason, you must either provide the frameworks in an agreed upon location, or store the frameworks in the bundle itself. Regardless of which solution you choose, you must make sure that the frameworks return the proper identification name for themselves, and that the application will look for these names. Luckily we can control this with the \c install_name_tool command-line tool. The \c install_name_tool works in two modes, \c -id and \c -change. The \c -id mode is for libraries and frameworks, and allows us to specify a new identification name. We use the \c -change mode to change the paths in the application. Let's test this out by copying the Qt frameworks into the Plug & Paint bundle. Looking at \c otool's output for the bundle, we can see that we must copy both the QtCore and QtGui frameworks into the bundle. We will assume that we are in the directory where we built the bundle. \snippet doc/src/snippets/code/doc_src_deployment.qdoc 39 First we create a \c Frameworks directory inside the bundle. This follows the Mac OS X application convention. We then copy the frameworks into the new directory. Since frameworks contain symbolic links, and we want to preserve them, we use the \c -R option. \snippet doc/src/snippets/code/doc_src_deployment.qdoc 40 Then we run \c install_name_tool to set the identification names for the frameworks. The first argument after \c -id is the new name, and the second argument is the framework which identification we wish to change. The text \c @executable_path is a special \c dyld variable telling \c dyld to start looking where the executable is located. The new names specifies that these frameworks will be located "one directory up and over" in the \c Frameworks directory. \snippet doc/src/snippets/code/doc_src_deployment.qdoc 41 Now, the dynamic linker knows where to look for QtCore and QtGui. Then we must make the application aware of the library locations as well using \c install_name_tool's \c -change mode. This basically comes down to string replacement, to match the identification names that we set for the frameworks. Finally, since the QtGui framework depends on QtCore, we must remember to change the reference for QtGui: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 42 After all this we can run \c otool again and see that the application will look in the right locations. Of course, the thing that makes the \l {tools/plugandpaint}{Plug & Paint} example interesting are its plugins. The basic steps we need to follow with plugins are: \list \o Put the plugins inside the bundle \o Make sure that the plugins use the correct library using the \c install_name_tool \o Make sure that the application knows where to get the plugins \endlist While we can put the plugins anywhere we want in the bundle, the best location to put them is under Contents/Plugins. When we built the Plug & Paint plugins, the \c DESTDIR variable in their \c .pro file put the plugins' \c .dylib files in a \c plugins subdirectory in the \c plugandpaint directory. So, in this example, all we need to do is move this directory: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 43 If we run \c otool on for example the \l {tools/plugandpaintplugins/basictools}{Basic Tools} plugin's \c .dylib file we get the following information. \snippet doc/src/snippets/code/doc_src_deployment.qdoc 44 Then we can see that the plugin links to the Qt frameworks it was built against. Since we want the plugins to use the framework in the application bundle we change them the same way as we did for the application. For example for the Basic Tools plugin: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 45 We must also modify the code in \c tools/plugandpaint/mainwindow.cpp to \l {QDir::cdUp()}{cdUp()} one directory since the plugins live in the bundle. Add the following code to the \c mainwindow.cpp file: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 46 \table \row \o \inlineimage deployment-mac-application.png \o The additional code in \c tools/plugandpaint/mainwindow.cpp also enables us to view the plugins in the Finder, as shown to the left. We can also add plugins extending Qt, for example adding SQL drivers or image formats. We just need to follow the directory structure outlined in plugin documentation, and make sure they are included in the QCoreApplication::libraryPaths(). Let's quickly do this with the image formats, following the approach from above. Copy Qt's image format plugins into the bundle: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 47 Use \c install_name_tool to link the plugins to the frameworks in the bundle: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 48 Then we update the source code in \c tools/plugandpaint/main.cpp to look for the the new plugins. After constructing the QApplication, we add the following code: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 49 First, we tell the application to only look for plugins in this directory. In our case, this is what we want since we only want to look for the plugins that we distribute with the bundle. If we were part of a bigger Qt installation we could have used QCoreApplication::addLibraryPath() instead. \endtable \warning When deploying plugins, and thus make changes to the source code, the default identification names are reset when rebuilding the application, and you must repeat the process of making your application link to the Qt frameworks in the bundle using \c install_name_tool. Now you should be able to move the application to another Mac OS X machine and run it without Qt installed. Alternatively, you can move your frameworks that live outside of the bundle to another directory and see if the application still runs. If you store the frameworks in another location than in the bundle, the technique of linking your application is similar; you must make sure that the application and the frameworks agree where to be looking for the Qt libraries as well as the plugins. \section2 Creating the Application Package When you are done linking your application to Qt, either statically or as frameworks, the application is ready to be distributed. Apple provides a fair bit of information about how to do this and instead of repeating it here, we recommend that you consult their \l {http://developer.apple.com/documentation/DeveloperTools/Conceptual/SoftwareDistribution/index.html}{software delivery} documentation. Although the process of deploying an application do have some pitfalls, once you know the various issues you can easily create packages that all your Mac OS X users will enjoy. \section1 Application Dependencies \section2 Qt Plugins Your application may also depend on one or more Qt plugins, such as the JPEG image format plugin or a SQL driver plugin. Be sure to distribute any Qt plugins that you need with your application, and note that each type of plugin should be located within a specific subdirectory (such as \c imageformats or \c sqldrivers) within your distribution directory, as described below. \note If you are deploying an application that uses QtWebKit to display HTML pages from the World Wide Web, you should include all text codec plugins to support as many HTML encodings possible. The search path for Qt plugins (as well as a few other paths) is hard-coded into the QtCore library. By default, the first plugin search path will be hard-coded as \c /path/to/Qt/plugins. But using pre-determined paths has certain disadvantages. For example, they may not exist on the target machine. For that reason you need to examine various alternatives to make sure that the Qt plugins are found: \list \o \l{qt-conf.html}{Using \c qt.conf}. This is the recommended approach since it provides the most flexibility. \o Using QApplication::addLibraryPath() or QApplication::setLibraryPaths(). \o Using a third party installation utility to change the hard-coded paths in the QtCore library. \endlist The \l{How to Create Qt Plugins} document outlines the issues you need to pay attention to when building and deploying plugins for Qt applications. \section2 Additional Libraries You can check which libraries your application is linking against by using the \c otool tool. To use \c otool, all you need to do is to run it like this: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 50 Unlike the deployment processes on \l {Deploying an Application on X11 Platforms}{X11} and \l {Deploying an Application on Windows}{Windows}, compiler specific libraries rarely have to be redistributed along with your application. But since Qt can be configured, built, and installed in several ways on Mac OS X, there are also several ways to deploy applications. Typically your goals help determine how you are going to deploy the application. The last sections describe a couple of things to keep in mind when you are deploying your application. \section2 Mac OS X Version Dependencies Qt 4.2 has been designed to be built and deployed on Mac OS X 10.3 up until the current version as of this writing, Mac OS X 10.4 and all their minor releases. Qt achieves this by using "weak linking." This means that Qt tests if a function added in newer versions of Mac OS X is available on the computer it is running on before it uses it. This results in getting access to newer features when running on newer versions of OS X while still remaining compatible on older versions. For more information about cross development issues on Mac OS X, see \l {http://developer.apple.com/documentation/DeveloperTools/Conceptual/cross_development/index.html}{Apple's Developer Website}. Since the linker is set to be compatible with all OS X version, you have to change the \c MACOSX_DEPLOYMENT_TARGET environment variable to get weak linking to work for your application. You can add: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 51 to your .pro file and qmake will take care of this for you. However, there is a bit of a wrinkle to keep in mind when your are deploying. Mac OS X 10.4 ("Tiger") ships GCC 4.0 as its default compiler. This is also the GCC compiler we use for building the binary Qt package. If you use GCC 4.0 to build your application, it will link against a dynamic libstdc++ that is only available on Mac OS X 10.4 and Mac OS X 10.3.9. The application will refuse to run on older versions of the operating system. For more information about C++ runtime environment, see \l {http://developer.apple.com/documentation/DeveloperTools/Conceptual/CppRuntimeEnv/index.html}{Apple's Developer Website} If you want to deploy to versions of Mac OS X earlier than 10.3.9, you must build with GCC 3.3 which is the default on Mac OS X 10.3. GCC 3.3 is also available on the Mac OS X 10.4 "Xcode Tools" CD and as a download for earlier versions of Mac OS X from Apple (\l {https://connect.apple.com/}{connect.apple.com}). You can use Apple's \c gcc_select(1) command line tool to switch the default complier on your system. \section3 Deploying Phonon Applications on Mac OS X \list \o If you build your Phonon application on Tiger, it will work on Tiger, Leopard and Panther. \o If you build your application on Leopard, it will \bold not work on Panther unless you rename the libraries with the following command after you have built your application: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 51a This command must be invoked in the directory where \c{libphonon_qt7.dylib} is located, usually in \c{yourapp.app/Contents/plugins/phonon_backend/}. \o The \l {macdeploy}{deployment tool} will perform this step for you. \o If you are using Leopard, but would like to build your application against Tiger, you can use: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 51b \endlist \section2 Architecture Dependencies The Qt for Mac OS X libraries, tools, and examples can be built "universal" (i.e. they run natively on both Intel and PowerPC machines). This is accomplished by passing \c -universal on the \c configure line of the source package, and requires that you use GCC 4.0.x. On PowerPC hardware you will need to pass the universal SDK as a command line argument to the Qt configure command. For example: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 52 From 4.1.1 the Qt binary package is already universal. If you want to create a binary that runs on older versions of PowerPC and x86, it is possible to build Qt for the PowerPC using GCC 3.3, and for x86 one using GCC 4.0, and use Apple's \c lipo(1) tool to stitch them together. This is beyond the scope of this document and is not something we have tried, but Apple documents it on their \l {http://developer.apple.com/documentation/}{developer website}. Once you have a universal Qt, \a qmake will generate makefiles that will build for its host architecture by default. If you want to build for a specific architecture, you can control this with the \c CONFIG line in your \c .pro file. Use \c CONFIG+=ppc for PowerPC, and \c CONFIG+=x86 for x86. If you desire both, simply add both to the \c CONFIG line. PowerPC users also need an SDK. For example: \snippet doc/src/snippets/code/doc_src_deployment.qdoc 53 Besides \c lipo, you can also check your binaries with the \c file(1) command line tool or the Finder. \section1 The Mac Deployment Tool \target macdeploy The Mac deployment tool can be found in QTDIR/bin/macdeployqt. It is designed to automate the process of creating a deployable application bundle that contains the Qt libraries as private frameworks. The mac deployment tool also deploys the Qt plugins, according to the following rules: \list \o Debug versions of the plugins are not deployed. \o The designer plugins are not deployed. \o The Image format plugins are always deployed. \o SQL driver plugins are deployed if the application uses the QtSql module. \o Script plugins are deployed if the application uses the QtScript module. \o The Phonon backend plugin is deployed if the application uses the \l{Phonon Module} {Phonon} module. \o The svg icon plugin is deployed if the application uses the QtSvg module. \o The accessibility plugin is always deployed. \o Accessibility for Qt3Support is deployed if the application uses the Qt3Support module. \endlist macdeployqt supports the following options: \list \o -no-plugins: Skip plugin deployment \o -dmg : Create a .dmg disk image \o -no-strip : Don't run 'strip' on the binaries \endlist */