I'd like to use a library (source codes from GH) in my JNI code. But the library depends on two other libraries (NTL and Boost) that are not available in Android NDK.
Now I am a bit confused and not sure if I understand correctly my following actions.
C++ code for Android is built into shared libraries (.so) for every platform (x86_64, armv7..). Does this mean that NTL, Boost and the lib I want to use must be compiled by me from source codes for these platforms too? If yes, how to do it correctly with cmake?
If I should build all the libs for specific platforms, how it is better to do, either as static libs (.a + headers) or as shared libs?
Do I really need to build NTL and Boost for all the platforms or I should do it just for the needed library?
Is Android.mk file required or can help with cmake? As I understand, it is used with "ndk-build" only.
Generally, if this sequence of actions is correct?
Build NTL for all platforms (.a + headers)
Build Boost for all platforms (.a + headers)
Build Library for all platforms (.so)
Add Library's .so-file as a dependency in CMakeLists for JNI project. (Do I still need dependent libs and headers or that dependencies will be incapsulated into lib?)
C++ code for Android is built into shared libraries (.so) for every platform (x86_64, armv7..). Does this mean that NTL, Boost and the lib I want to use must be compiled by me from source codes for these platforms too? If yes, how to do it correctly with cmake?
Yes, you'll need to build those libraries from source (or find a binary distribution for Android) if you want to use those libraries in your application. As for how to do that, you'll have to wait for someone else to answer or try Googling it. There are a handful of "how to build X for Android" tutorials out there, but I don't know if you'll find many for CMake since CMake is pretty new for Android.
If I should build all the libs for specific platforms, how it is better to do, either as static libs (.a + headers) or as shared libs?
That mostly depends on how many shared libraries you're building for your app. The ideal model for an app is to use a single shared library in your app and statically link in all of your dependencies (going to avoid linker bugs on old versions of Android, and will make your app as small as possible). If you have multiple shared libraries for your code, you'll need to use shared libraries for your dependencies to avoid ODR issues.
Do I really need to build NTL and Boost for all the platforms or I should do it just for the needed library?
You'll need to do it for any platform you need to use those libraries on.
Is Android.mk file required or can help with cmake? As I understand, it is used with "ndk-build" only.
CMake and ndk-build should both work, but you might have an easier time finding porting instructions for ndk-build due to CMake's relative youth in Android.
Related
I have an issue where I'm developing on one system where Boost is installed on:
/usr/include
/usr/lib
On a system I will deploy this on, the libboost libraries are at:
/nfs/mount/boost
/nfs/mount/lib
And I can't go about changing every system I deploy on to install libboost in the same place.
Is there a way to either:
include libboost as part of the binary executable such that loading from the system lib paths is not needed.
make the executable search for different directories when trying to load to libboost?
I'm using g++ 8
Sounds like you need a more sophisticated build environment.
I'm not sure what you mean here:
include libboost as part of the binary executable such that linking is not needed
Linking is not something you can skip. If you are trying to avoid distributing .dll/.so files with your executable, you need to avoid using the portions of the boost library that require compilation of the boost binaries (i.e. those not listed here https://www.boost.org/doc/libs/1_62_0/more/getting_started/windows.html#header-only-libraries).
Of course, that is not often the case. So...
make the executable search for different directories when trying to link to libboost?
This will never work reliably as you scale and is a nightmare in the CI world. This is where package managers such as conan (https://conan.io/) come to save the day. Delegating the package management to a third-party is the most reliable way of getting your code to build across multiple environments/platforms.
Also, building your executable and distributing it are separate operations. Any dynamically linked libraries will need to be discoverable on the system path at runtime.
Just now I noted that the MinGW Toolchain that comes with the default Qt installation, at least Qt 5.14, comes with a lib directory with libQt5*.a files. Are those files static libraries?
I think so because:
They have a size similar to the ones that I get when I compile Qt statically for release.
$file ./libQt5Core.a outputs ./libQt5Core.a: current ar archive, which is the same that it outputs for the statically compiled ones.
If indeed they're static libraries, how can I tell QMake (for example editing the .pro file) to link to those instead of linking to the shared ones?
Are those files static libraries?
No. They are not static. Qt’s default online installer provides only shared libraries. That *.a files are so-called import libraries.
Import library is an .a or .lib library, but it only contains bit of information needed to tell the linker/OS how your program interacts with the dll’s.
If you need Qt static windows build for some reasons, you have some options:
HARD Build whole Qt (or needed modules) by yourself.
EASY Use vcpkg: vcpkg install qt5:x64-windows-static
With vcpkg you can create custom MinGW triplet if you need MinGW for some reason. But I suggest you stick with MSVC.
I have already create a static and dynamic library for c++ in ubuntu and I wanna know how I'm gonna distribute this library for others, so they can install it and use it with the same plateform (linux).
I have the file (.h) and files (.so and .a) for the librairy.
Please your help
There're so many ways to share your libs.
One could be creating a github project, with CMakeLists (or other buid system), which would be able to build lib on other people platforms.
Don't forget about the license (:
I have used Boost library (particularly, Boost filesystem) for my project in Visual Studio C++ (9.0 version). This Boost library is installed in C drive. My project is already finished, therefore, I have corresponding Debug and Release.
Now, I want to include particular DLLs of Boost filesystem to Debug and Release, so my first question is which DLLs do I need to copy? It shows many DLLs for filesystem. Ones which I tried are "boost_filesystem-vc90-mt-1_40.dll" and "boost_filesystem-vc90-mt-gd-1_40.dll". Since, I got an idea from the "vc90", I think it means for Visual Studio 9.0?
Second question, even if I don't include these Boost filesystem DLLs to Debug or Release, my EXE file of the project is still working? Can anyone please explain why is this happening? I tried to check if I have Boost Library path somehow in my Environment Variables, but, unfortunately I could not find them there. Thanks a lot.
P.S.: I am a beginner in both C++ and Boost. And, another library that I am using is OpenCV, for which I already copied corresponding DLLs (core, highgui and imgproc) to Debug and Release and this one is working perfectly.
You have 2 options:
a. Deploy Boost DLLs together with your binary. Answering your question - of course if your project depends on Boost, the libraries are required at runtime, your binary will not start without them.
If you deploy Release configuration of your project, then you need boost_filesystem-vc90-mt-1_40.dll library. In general you can just see all the dependencies of your binary using Dependency Walker tool.
b. Link Boost libraries statically, in this case your binary will not have runtime dependency on Boost DLLs, so you don't have to deploy them
I'm developing an application using Qt and OpenGL, and found it necessary to download the GLM library. It's a header-only library, so I don't need to link to anything. I want this application to build on any system that has the correct libraries installed. My problem is that I don't know where to put GLM so that the system can find it without adding a specific path to the project's .pro file. The .pro file is part of my git repository, which is how the source is distributed to other systems like Linux, etc. So I don't want this file to specify the exact location of GLM because other systems could have it in other places.
I'm currently developing on Windows, compiling in Qt Creator using Visual C++ 2010. I read from MSDN that #include <file> searches using the INCLUDE environment variable, so I tried to add the path to glm.hpp to INCLUDE, but QtCreator's build environment for this project seems to overwrite INCLUDE. So I appended the path to GLM to the new INCLUDE from within QtCreator's Projects tab, but my code still can't find glm.hpp.
In general, how can I add an external library to my system such that my compiler will be able to find it without specifying the exact location in a project file that's distributed to other systems?
What you need is a build system with the power to search the system for the libraries you request, and be able to do so on any platform. One such build system is cmake, and it is the most widely used build system. In essence, cmake allows you to write a build script in which you can specify all the things you normally specify when creating a project in Qt Creator or Visual Studio, like the list of source files, grouped by targets to compile (libraries, executables, etc.), the relative paths to the headers, and libraries to include for linking and for include-paths, amongst many more things. For libraries that are installed on the system, there is a function, called find_package() (part of cmake script commands), that will find out if the library is installed and where to find its lib files and headers (storing those paths as cache strings that you can specify on the targets and such). It usually works great, as long as the libraries are not installed in weird places. The way it works is that when you run cmake, it will generate a build script/configuration for almost any platform you specify, and then you use that to compile your code. It can generate makefiles (Unix-like or windows), CodeBlocks project files, Visual Studio project files, etc.. And many IDEs also have native support for cmake projects.
I wish I could recommend an alternative, just to sound less biased for cmake, but I haven't heard of any that truly compare to it, especially for the purpose of locating external dependencies and working with different platforms. I would guess Boost.Build is decent too.