I want to move my current project to C++11. The code all compiles using clang++ -std=c++0x. That is the easy part :-) . The difficult part is dealing with external libraries. One cannot rely on linking one's C++11 objects with external libraries that were not compiled with c++11 (see http://gcc.gnu.org/wiki/Cxx11AbiCompatibility). Boost, for example, certainly needs re-building (Why can't clang with libc++ in c++0x mode link this boost::program_options example?). I have source for all of the external libraries I use, so I can (with some pain) theoretically re-build these libs with C++11. However, that still leaves me with some problems:
Developing in a mixed C++03/C++11 environment: I have some old projects using C++03 that require occasional maintenance. Of course, I'll want to link these with existing versions of external libraries. But for my current (and new) projects, I want to link with my re-built C++11 versions of the libraries. How do I organise my development environments (currently Ubuntu 12.04 and Mac OS X 10.7) to cope with this?
I'm assuming that this problem will be faced by many developers. It's not going to go away, but I haven't found a recommended and generally approved solution.
Deployment: Currently, I deploy to Ubuntu 12.04 LTS servers in the cloud. Experience leads one to depend (where possible) on the standard packages (e.g. libboost) available with the linux distribution. If I move my current project to c++11, my understanding is that I will have to build my own versions of the external libraries I use. My guess is that at some point this will change, and their will be 'standard' versions of library packages with C++11 compatibility. Does anyone have any idea when one might expect that to happen? And presumably this will also require a standard solution to the problem mentioned above - concurrent existence of C++03 libs and C++11 libs on the same platform.
I am hoping that I've missed something basic so that these perceived problems disappear in a puff of appropriate information! Am I trying to move to C++11 too soon?
Update(2013-09-11): Related discussion for macports: https://lists.macosforge.org/pipermail/macports-users/2013-September/033383.html
You should use your configure toolchain (e.g. autotools) to "properly" configure your build for your target deployment. Your configuration tests should check for ABI compatible C++11 binaries and instruct the linker to use them first if detected. If not, optionally fail or fallback to a C++03 build.
As for C++11 3rd part libraries installed in a separate parallel directory tree, this isn't strictly necessary. Library versioning has been around for a long time and allows you to place different versions side by side on the system or wherever you'd like, again based on configure.
This might seem messy, but configure toolchains were designed to handle these messes.
Related
I've written a fairly substantial C++11 library, and I'm planning to allow download of pre-compiled versions from my website.
So I've set up an automated build which compiles the library with clang and make it available for download, but this has exposed a problem: if I try to use the clang-compiled library with GCC I get undefined references (mainly related to std::string). I think this is related to the GCC dual-ABI changes in GCC 5.1 but I'm not sure how to fix it.
My question is, what flags should I set, or practices should I follow to make a C++ library compatible with both clang and GCC?
Or should I give up and compile two separate libraries?
As already mentioned in several places (eg. here) libc++ is not fully binary compatible with libstdc++.
There are several options, but some of them are somewhat not-so-straightforward.
Compile two separate libraries - always working solution.
Remove incompatible containers from your interface (eg. std::string) - but this might be lot of work and sometimes not a good idea.
Instruct your library GCC users that you link with libc++ and that they need to do the same, basic steps here. But I guess most GCC users do not want to do this.
Use clang with libstdc++ using -stdlib=libstdc++ flag to be compatible with libstdc++ (as suggested in other answer). This solution might be harder to setup on some platforms though.
I would suggest as already mentioned in comments to go with option 1.
There are several options:
Don't distribute it in binary form. Instead, make it easy to build everywhere (e.g. by using CMake, or autotools or ...)
Make it header only. This is by far the simplest solution but might not be what you want. It only really makes sense for templated code, and incurs a heavy impact on compile-time performance of your library.
Tell people to link with libstdc++ when using Clang and your library. Suboptimal solution (I for one like to check my code against libc++ as well as libstdc++), but (virtually) every Linux user has libstdc++ installed anyway. Make sure to pick a slightly older version (the one shipped in the latest Debian Stable distro is a good choice), because newer versions might introduce new symbols olders versions are missing. New versions should be ABI compatible anyway.
Note the situation for Visual Studio users is even worse, where every single compiler release mandates a new binary because they guarantee absolutely nothing with respect to the C++ libraries' or compiler's ABI.
Another option is for your shared library to not expose any C++ standard library types in its interface. And have a header file supplied with your shared library that converts std::string to types consumed by your library, such as struct my_string_span { char const *begin, *end; }; and other standard containers as necessary.
I'm working on a large project, cross platform between Linux and OSx.
I'm would like to include boost functionalities, but I don't want to force all the the collaborators to install on their machine (with totally different environment) all the boost libraries.
If I compile boost on my machine, and put the static libraries inside the repository, which problem could I face? Can my colleagues use the same static libraries on their environment?
AFAIK there will be difference. The static libraries are not the same on OS X and Linux. Also compilation depends on toolset (see boost guide). And there can be issues if IDEs are different.
However you can compile both versions from one platform (see Cross-compilation) and put them in repository, but putting binary objects is not the best idea (even if they don't depend on platform).
I think you could try to compile and link boost on different platforms, maybe you will succeed, but you can not cover all the scenarios for sure. It is better to create boost compilation script and tell everyone to use it.
C++ Buildsystem with ability to compile dependencies beforehand
Java has Maven which is a pleasure to work with, simply specifying dependencies that are already compiled, and deposited to Mavens standard directory, meaning that the location of the dependencies is standardized as opposed to the often used way of having multiple locations (give me a break, like anyone remembers the default installed directories for particular deps) of C/C++ dependencies.
It is massively unproductive for every individual developer having to, more often than not, find, read about, get familiar with the configure options/build, and finally compile for every dependency to simply make a build of a project.
What is the theoretical reason this has not been implemented?
Why would it be difficult to provide packages of the following options with a maven-like declaration format?
version
platform (windows, linux)
src/dev/bin
shared/static
equivalent set of Boost ABI options when applicable
Having to manually go to websites and search out dependencies in the year 2013 for the oldest major programming language is absurd.
There aren't any theoretical reasons. There are a great many practical reasons. There are just too many different ways of handling things in the C++ world to easily standardize on a dependency system:
Implementation differences - C++ is a complicated language, and different implementations have historically varied in how well they support it (how well they can correctly handle various moderate to advanced C++ code). So there's no guarantee that a library could be built in a particular implementation.
Platform differences - Some platforms may not support exceptions. There are different implementations of the standard library, with various pros and cons. Unlike Java's standardized library, Windows and POSIX APIs can be quite different. The filesystem isn't even a part of Standard C++.
Compilation differences - Static or shared? Debug or production build? Enable optional dependencies or not? Unlike Java, which has very stable bytecode, C++'s lack of a standard ABI means that code may not link properly, even if built for the same platform by the same compiler.
Build system differences - Makefiles? (If so, GNU Make, or something else?) Autotools? CMake? Visual Studio project files? Something else?
Historical concerns - Because of C's and C++'s age, popular libraries like zlib predate build systems like Maven by quite a bit. Why should zlib switch to some hypothetical C++ build system when what it's doing works? How can a newer, higher-level library switch to some hypothetical build system if it depends on libraries like zlib?
Two additional factors complicate things:
In Linux, the distro packaging systems do provide standardized repositories of development library headers binaries, with (generally) standardized ABIs and an easy way of specifying a project's build dependencies. The existence of these (platform-specific) solutions reduces the impetus for a cross-platform solution.
With all of these complicating factors and pre-existing approaches, any attempt to establish a standard build system is going to run into the problem described in XKCD's "Standards":
Situation: There are 14 competing standards.
"14? Riculous! We need to develop one universal standard that covers everyone's use cases."
Soon: There are 15 competing standards.
With all of that said:
There is some hope for the future. For example, CMake seems to be gradually replacing other build systems. Some of the Boost developers have started Ryppl, an attempt to do what you're describing.
(also posted in linked question)
Right now I'm working on a tool able to automatically install all dependencies of a C/C++ app with exact version requirement :
compiler
libs
tools (cmake, autotools)
Right now it works, for my app. (Installing UnitTest++, Boost, Wt, sqlite, cmake all in correct order)
The tool, named «C++ Version Manager» (inspired by the excellent ruby version manager) is coded in bash and hosted on github : https://github.com/Offirmo/cvm
Any advices and suggestions are welcomed.
well, first off a system that resolves all the dependencies doesn't makes you productive by default, potentially it can make you even less productive.
Regarding the differences between languages I would say that in Java you have packages, which are handy when you have to organize and give a limited horizon to your code, in C++ you don't have an equivalent concept.
In C++ all the libraries that can solve a symbol are good enough for the compiler, the only real requirement for a library is to have a certain ABI and to solve the required symbols, there are no automated ways that you can work to pick the right library, also solving a symbol it's just a matter of linking your function to the actual implementation, this doesn't even grant you that a correct linking phase will make your app work.
To this you can add important variables such as the library version, different implementations of the same library and different libraries with the same methods name.
An example is the Mesa library VS the opengl lib from the official drivers, or whatever lib you want that offers multiple releases and each one can solve all the symbols but probably there is a release that is more mature than the others and you can ask a compiler to pick the right one because they are all the same for its own purposes .
I'm writing a cross-platform application which is not GNU GPL compatible. The major problem I'm currently facing is that the application is linked dynamically with glibc and libstdc++, and almost every new major update to the libraries are not backwards compatible. Hence, random crashes are seen in my application.
As a workaround, I distribute binaries of my application compiled on several different systems (with different C/C++ runtime versions). But I want to do without this. So my question is, keeping licensing and everything in mind, can I link against glibc and libstdc++ statically? Also, will this cause issues with rtld?
You don't need to.
Copy the original libraries you linked against to a directory (../lib in this example) in your application folder.
Like:
my_app_install_path
.bin
lib
documentation
Rename you app for something like app.bin. Substitute your app for a little shell script that sets the enviroment variable LD_LIBRARY_PATH to the library path (and concatenate the previous LD_LIBRARY_PATH contents, if any). Now ld should be able to find the dynamic libraries you linked against and you don't need to compile them statically to your executable.
Remember to comply with the LGPL adding the given attribution to the libraries and pointing in the documentation where the source can be downloaded.
glibc is under the LGPL. Under section 6. of LGPL 2.1, you can distribute your program linked to the library provided you comply with one of five options. The first is to provide the source code of the library, along with the object code (source is optional, not required) of your own program, so it can be relinked with the library. You can alternatively provide a written offer of the same. Your own code does not have to be under the LGPL, and you don't have to release source.
libstdc++ is under the GPL, but with a major exception. You can basically just distribute under the license of your choice without providing source for either your own code or libstdc++. The only condition is that you compile normally, without e.g. proprietary modifications or plugins to GCC.
IANAL, and you should consider consulting one if you need real legal advice.
Specifying the option -static-libgcc to the linker would cause it to link against a static version of the C library, if available on the system. Otherwise it is ignored.
I must question what the heck you are doing with the poor library functions?
I have some cross platform software as well. It runs fine on Linux systems of all sorts. Build with the oldest version of software that you want to support. The glibc and libstdc++ libraries are really very backward compatible.
I have built on CentOS 4 and run it on RHEL 6 beta. No problems.
I can build on stable Debian and run it on testing.
Now, I do sometimes have trouble with some libraries if I try to build on, say old Debian and try to run it on CentOS 5.4. That is usually due to distribution configuration choices that are different, like choosing threading or non-threading.
Does anyone have any experience with running C++ applications that use the boost libraries on uclibc-based systems? Is it even possible? Which C++ standard library would you use? Is uclibc++ usable with boost?
We use Boost together with GCC 2.95.3, libstdc++ and STLport on an ARMv4 platform running uClinux. Some parts of Boost are not compatible with GCC 2.x but the ones that are works well in our particular case. The libraries that we use the most are date_time, bind, function, tuple and thread.
Some of the libraries we had issues with were lambda, shared_pointer and format. These issues were most likely caused by our version of GCC since it has problems when you have too many includes or deep levels of template structures.
If possible I would recommend you to run the boost test suite with your particular toolchain to ensure compatibility. At the very least you could compile a native toolchain in order to ensure that your library versions are compatible.
We have not used uClibc++ because that is not what our toolchain provider recommends so I cannot comment on that particular combination.
We are using many of the Boost libraries (thread, filesystem, signals, function, bind, any, asio, smart_ptr, tuple) on an Arcom Vulcan which is admittedly pretty powerful for an embedded device (64M RAM, 533MHz XScale). Everything works beautifully.
GCC 3.4 but we're not using uclib++ (Arcom provides a toolchain which includes libstd++).
Many embedded devices will happily run many of the Boost libraries, assuming decent compiler support. Just take care with usage. The Boost libraries raise the level of abstraction and it can be easy to use more resources than you think.
I googled "uclibc stlport". It seems there are at least a few versions of uclibc for which stlport can be compiled (see this).
Given that, i'd say Boost is just a few compilation steps away. I have read a message by David Abrahams (who is an active member of the boost community) that says that Boost does not depend directly on the used libc. But some libraries may still cause problems, Boost.Python for instance, since it depends on something else (Python in my example) that might be difficult to compile with uclibc.
Hope this helps
I have not tried but I don't know anything about uclibc that would prevent Boost from working.
Try it and see what happens, I would say.
Yes you can use boost with uclibc.
I tried this with boost 1.45 & uclibc on ARM9260
Use fresh OpenEmbedded
Configure it to use Angstrom
Configure Angstrom to use uclibc
make boost - bitbake boost