My Dockerfile uses FROM gcc:5, runs apt-get update and installs CMake v3.9 via wget. My top-level CMakeLists.txt has set(CMAKE_CXX_STANDARD 11) but that doesn't seem to convince gcc to compile using C++11 as I get the following error:
/ifeature.hpp:51:42: error: 'shared_ptr' in namespace 'std' does not name a template type
virtual float compare(const std::shared_ptr<IFeature>& feature) const = 0
^
I tried adding variations of set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11") with libstdc++, gnuc++11 etc. but the compiler does not recognize them. I also tried add_compile_options(-std=c++11) to no avail. I also tried apt-get upgrade gcc but that didn't help either.
What am I missing here?
For future reference and in the interest of anyone who may encounter this page, I should update that just as #some-programmer-dude mentioned in the comment, simply including <memory> resolved the error shown in my question.
Given that my original development environment was different from the one in my Docker image and the code was written with different compiler, I had to resolve a few similar issues (missing include files) to satisfy gcc-v5 requirements.
Related
I would like to use Open3d in LAMMPS. Open3D details how to find the pre-installed Open3D package using cmake.
Using the above, I have written a cmake file that I believe LAMMPS uses during its build stage to find packages and link them. Curiously, the line
target_link_libraries(lammps PRIVATE Open3D::Open3D)
appears to cause the compiler to find errors in the LAMMPS src code, i.e.,
/home/USER/lammps/src/fmtlib_format.cpp:58:51: error: duplicate explicit instantiation of ‘struct fmt::v7_lmp::detail::basic_data<void>’ [-fpermissive]
58 | template struct FMT_INSTANTIATION_DEF_API detail::basic_data<void>;
If I comment the target_link_libraries statement, the code compiles just fine (but doesn't link to Open3D). Apologies for being unable to provide a MWE as I do not know how to replicate this behaviour in a simple manner. Could you please explain to me why the target_link_libraries command causes the compiler to find errors in the LAMMPS src code and provide a solution to prevent this from occuring? If relevant I am using Clion 2021.1.2 and
CMAKE_CXX_COMPILER_VERSION = 9.3.0
CMAKE_CXX_STANDARD = 11
CMAKE_VERSION = 3.19.2
I am new to CMake projects and I want to use the file system library in my project. I am running Ubuntu 18.04 with GCC 8.2 and CMake 3.13. In order to achieve this I tried two options:
Option 1
cmake_minimum_required(VERSION 3.13)
project(TheFsProject)
set(CMAKE_CXX_STANDARD 17)
set(CMAKE_CXX_FLAGS "-std=c++17 -lstdc++fs")
This does not help as the compiler still cannot find the file system
library during compile time.
Option 2 (copied from:
https://www.scivision.co/cmake-cpp-17-filesystem/)
make_minimum_required(VERSION 3.13)
project(TheFsProject)
set(CMAKE_CXX_STANDARD_REQUIRED ON)
set(CMAKE_REQUIRED_FLAGS -std=c++17)
include(CheckCXXSymbolExists)
CHECK_CXX_SYMBOL_EXISTS(std::filesystem::path::preferred_separator cxx17fs)
if(cxx17fs)
add_executable(TheFsProject main.cpp)
set_property(TARGET TheFsProject PROPERTY CXX_STANDARD 17)
endif()
This does not help either as I get a CMake error which I don't
understand.
(CHECK_CXX_SYMBOL_EXISTS):
CHECK_CXX_SYMBOL_EXISTS Macro invoked with incorrect arguments for macro named: CHECK_CXX_SYMBOL_EXISTS
I feel out of my depth on this topic which is why I came here. I don't mind putting extra work into finding out more but I don't know anymore where to look. Any help would be appreciated!
EDIT 1
Thanks for the replies so far! I made Option 3 based on your feedback:
cmake_minimum_required(VERSION 3.13)
project(TheFsProject)
set(CMAKE_CXX_STANDARD 17)
set(CMAKE_CXX_STANDARD_REQUIRED ON)
add_executable(TheFsProject main.cpp)
target_link_libraries(TheFsProject stdc++fs)
Sadly it doesn't fix my problem. It still issues an error during compilation that it can't find the compilation header.
EDIT 2
Thanks for all the replies so far. All of these help. I tried Ashkan his answer last (because it seemed intimidating). This one returns
Compiler is missing file system capabilities.
so I'm guessing something is wrong on that end. This is useful in the sense that I now know it's probably not due to my CMake file. I now have to find out why the compiler does support the file system header though...
EDIT 3
Strictly speaking this question is answered because my question is about the CMake file. I am going to mark Ashkan his answer as the solution simply because it produced the next step in my troubleshooting search. If I could I would also mark lubgr his answer because I think that's a really good answer as well. Thanks everyone!
Gcc 8.2. comes with <filesystem>, so there is no need to investigate with regard to the availability. Next, option 1 is sufficient, but needs a fix:
set(CMAKE_CXX_STANDARD 17) # no need to manually adjust the CXXFLAGS
add_executable(yourExecutable yourSourceFile.cpp)
target_link_libraries(yourExecutable stdc++fs)
This should result in compiling the sources with -std=c++17 or -std=gnu++17 and adding -lstdc++fs when linking.
Edit: Note that as #Ashkan has pointed out in the comments, setting CMAKE_CXX_STANDARD_REQUIRED to true results in an immediate error at configure time if C++17 isn't supported by the compiler, instead of a compilation error (due to the missing <filesystem> header) or at link time (due to the missing shared library). This might be desirable.
Besides from #lubgr's answer. I think a more complete way is to also do try_compile to see that you can actually use the filesystem header. This in my opinion is better because some compilers are not supporting std::filesystem yet. Also in gcc 7.x you have the filesystem under experimental namespace. This way you can have a separate try_compile in the else clause and detect that.
Here is the related cmake for it
# set everything up for c++ 17 features
set(CMAKE_CXX_STANDARD 17)
# Don't add this line if you will try_compile with boost.
set(CMAKE_CXX_STANDARD_REQUIRED ON)
# test that filesystem header actually is there and works
try_compile(HAS_FS "${CMAKE_BINARY_DIR}/temp"
"${CMAKE_SOURCE_DIR}/tests/has_filesystem.cc"
CMAKE_FLAGS -DCMAKE_CXX_STANDARD=17 -DCMAKE_CXX_STANDARD_REQUIRED=ON
LINK_LIBRARIES stdc++fs)
if(HAS_FS)
message(STATUS "Compiler has filesystem support")
else()
# .... You could also try searching for boost::filesystem here.
message(FATAL_ERROR "Compiler is missing filesystem capabilities")
endif(HAS_FS)
The file tests/has_filesystem.cc is very simple
#include <filesystem>
namespace fs = std::filesystem;
int main()
{
fs::path aPath {"../"};
return 0;
}
You could in your else clause try_compile for boost::filesystem and pass a directive that can be used in your source file where you decide if you want to use c++17 filesystem or boost.
CHECK_CXX_SYMBOL_EXISTS takes three arguments, not two:
include(CheckCXXSymbolExists)
check_cxx_symbol_exists(std::filesystem::path::preferred_separator filesystem cxx17fs)
You forgot to tell CMake where to look for the symbols (the header that declares them).
I have found a case when try_compile was not enough: when using the Intel C++ compiler (icpc (ICC) 19.1.1.216 20200306) in C++17 mode running on MacOS "Mojave" 10.14.6. The test program recommended by #Ashkan compiled without errors, and it even ran. However, my code uses fs::path::filename() at one point and that resulted in a runtime linker error (dyld: lazy symbol binding failed: Symbol not found:). In other words: the header is there, the implementation apparently isn't (?). I didn't investigate this any further.
The solution is to use try_run instead of try_compile and (in my case) fall back to boost::filesystem if std::filesystem is not yet supported.
Here is the relevant CMake code section:
try_run(RUNS_WITH_STDFS COMPILES_WITH_STDFS
"${CMAKE_BINARY_DIR}/try"
"${CMAKE_SOURCE_DIR}/cmake/has_stdfs.cc"
CMAKE_FLAGS CMAKE_CXX_STANDARD=17 CMAKE_CXX_STANDARD_REQUIRED=ON
)
if (RUNS_WITH_STDFS STREQUAL "FAILED_TO_RUN")
message(STATUS "Using boost::filesystem instead of std::filesystem")
set(_boost_components ${_boost_components} filesystem system)
add_definitions(-DUSE_BOOST_FILESYSTEM)
else()
message(STATUS "std::filesystem supported")
endif()
Note that the variable RUNS_WITH_STDFS is not set to NO in case of failure but to "FAILED_TO_RUN" which is not interpreted as a FALSE Boolean (see CMake if() docs:
if() True if the constant is 1, ON, YES, TRUE, Y, or a
non-zero number. False if the constant is 0, OFF, NO, FALSE, N,
IGNORE, NOTFOUND, the empty string, or ends in the suffix -NOTFOUND.
so I had to string-compare its value.
The little test program also changed a bit compared to #Ashkan's solution:
// == PROGRAM has_stdfs.cc ==
// Check if std::filesystem is available
// Source: https://stackoverflow.com/a/54290906
// with modifications
#include <filesystem>
namespace fs = std::filesystem;
int main(int argc, char* argv[]) {
fs::path somepath{ "dir1/dir2/filename.txt" };
auto fname = somepath.filename();
return 0;
}
I'm trying to compile the EOS blockchain/smart contract project on GitHub on Ubuntu 14.04:
https://github.com/EOSIO/eos
After getting Clang 4.0 to install, installing build_essentials, and upgrading CMake to 3.5, I was able to run the build process without any missing dependencies. However, now I get the errors shown below when I build the EOS source. This seems to me like another general issue with the configuration of the tools on my system since many people are able to compile the EOS code, although usually on Ubuntu 14.04.
Can anyone tell by looking at the errors I'm getting what tool or library I need to install or upgrade?
In file included from /usr/lib/llvm-4.0/include/clang/AST/Decl.h:31:
/usr/lib/llvm-4.0/include/llvm/Support/TrailingObjects.h:259:33: error: 'BaseTy' does not refer to a value
static_assert(LLVM_IS_FINAL(BaseTy), "BaseTy must be final.");
^
/usr/lib/llvm-4.0/include/llvm/Support/TrailingObjects.h:233:20: note: declared here
template <typename BaseTy, typename... TrailingTys>
^
/usr/lib/llvm-4.0/include/llvm/Support/TrailingObjects.h:259:19: error: expected expression
static_assert(LLVM_IS_FINAL(BaseTy), "BaseTy must be final.");
^
/usr/lib/llvm-4.0/include/llvm/Support/type_traits.h:104:45: note: expanded from macro 'LLVM_IS_FINAL'
#define LLVM_IS_FINAL(Ty) std::is_final<Ty>()
^
Linking CXX executable codegen
/home/robert/Documents/GitHub/eos/programs/launcher/main.cpp:405:18: error: no template named 'underlying_type_t' in namespace 'std'; did you mean
'underlying_type'?
using T = std::underlying_type_t <enum_type>;
~~~~~^~~~~~~~~~~~~~~~~
underlying_type
/usr/bin/../lib/gcc/x86_64-linux-gnu/4.8/../../../../include/c++/4.8/type_traits:1855:12: note: 'underlying_type' declared here
struct underlying_type
^
/home/robert/Documents/GitHub/eos/programs/launcher/main.cpp:435:17: error: no member named 'put_time' in namespace 'std'
dstrm << std::put_time(std::localtime(&now_c), "%Y_%m_%d_%H_%M_%S");
~~~~~^
[ 64%] Building CXX object libraries/chain/CMakeFiles/eos_chain.dir/chain_controller.cpp.o
/home/robert/Documents/GitHub/eos/programs/launcher/main.cpp:406:39: error: no matching conversion for static_cast from 'allowed_connection' to 'T'
(aka 'underlying_type<allowed_connection>')
return lhs = static_cast<enum_type>(static_cast<T>(lhs) | static_cast<T>(rhs));
^~~~~~~~~~~~~~~~~~~
The Missing _t alias names look like you’re having issues with C++14. The header paths in the error messages look like you’re using the standard library from GCC 4.8 (the default compiler on Ubuntu 14.04), which is simply too old.
I can see two solutions:
Switch from GCC’s libstdc++ to an up-to-date version of LLVM’s libc++. I’m not familiar enough with Ubuntu to tell you how to install it. For the compilation of EOSIO you must pass the -stdlib=libc++ option to Clang to switch to the different stdlib. EOSIO looks like it’s using CMake, so you have to include -DCMAKE_CXX_FLAGS=-stdlib=libc++ in your CMake command line.
Use the Toolchain test builds PPA to install a newer GCC and and libstdc++ in addition to your system’s default one. For Ubuntu 14.04 GCC 7.2.0 is the latest version available, which is perfectly C++14 capable. Add the PPA to your package sources and then do a:
sudo apt-get install gcc-7 g++-7
This installs both the GCC C compiler and C++ compiler along with the stdlib. Your default compiler is still going to be the old GCC 4.8, so you’ll have to tell CMake about the newer versions:
-DCMAKE_CXX_COMPILER=g++-7 -DCMAKE_C_COMPILER=gcc-7
Note that now you compile EOSIO with GCC (and the new stdlib) instead of Clang. Instructing Clang to use a specific version of libstdc++ should be possible, but I don’t know how.
Official support is for Ubuntu 16.10. Consider upgrading.
(EDITED: I mistakenly said 14.10)
Source: https://github.com/EOSIO/eos/wiki/Local-Environment#211-ubuntu-1610
First of all I am complete new to C++, so if you know the answer please be patient with me ;). Here my problem:
I wanna solve an IP with Gurobi in a C++ Code. The Code itself seems fine since there are no expression marked as errors. However when I run the Code I get the following error report:
Undefined symbols for architecture x86_64:
and many lines like that:
"GRBLinExpr::GRBLinExpr(GRBVar, double)", referenced from:
bridge_problem::max_flow_lp(time_expanded_network&, lemon::ListDigraph&, lemon::DigraphExtender<lemon::ListDigraphBase>::ArcMap<int>&, lemon::DigraphExtender<lemon::ListDigraphBase>::ArcMap<int>&, lemon::DigraphExtender<lemon::ListDigraphBase>::NodeMap<int>&) in bridge_problem.cpp.o
I suppose that the mistake is in my CMakeList.txt file. This file was automatically created since I am using Clion and for including gurobi I entered those additional lines:
include_directories(/Library/gurobi604/mac64/include)
link_directories(/Library/gurobi604/mac64/lib/libgurobi_c++.a)
link_directories(/Library/gurobi604/mac64/lib/libgurobi60.so)
Any help is greatly appreciated and if you need any more information just let me know. (In case it is important I using a mac).
EDIT: I changed my make code since I found this one:
https://github.com/joschu/trajopt/blob/master/cmake/modules/FindGUROBI.cmake
I changed the version since I have gurobi604 but it still does not work. My new error message is:
fatal error: 'gurobi_c++.h' file not found #include "gurobi_c++.h"
I don't get it since I thought by
find_path(GUROBI_INCLUDE_DIR
NAMES gurobi_c++.h
PATHS "$ENV{GUROBI_HOME}/include"
"/Library/gurobi604/mac64/include"
"C:\\libs\\gurobi604\\include"
)
that should be easy to find. Any suggestion?
From the small excerpt of your CMakeLists.txt, I expect you should be using target_link_libraries rather than link_directories.
I'd normally recommend linking to the static version of any library rather than the shared if possible (i.e. in this case prefer "libgurobi_c++.a" over "libgurobi60.so" assuming they're the same library, just compiled differently).
So, if your exe is called MyExe, you could do:
target_link_libraries(MyExe /Library/gurobi604/mac64/lib/libgurobi_c++.a)
Also, it's almost always best to avoid specifying hard-coded paths in your CMakeLists.txt. Although your copy of "libgurobi_c++.a" lives in "/Library/gurobi604/mac64/lib/", that won't be the case for other users, or on different platforms.
You can avoid this by having CMake "find" the library, for example by calling find_library:
find_library(Gurobi NAMES gurobi_c++)
if(NOT Gurobi)
message(FATAL_ERROR "Failed to find Gurobi lib. Try setting CMAKE_PREFIX_PATH")
endif()
target_link_libraries(MyExe ${Gurobi})
Then, when you run CMake, you just need to tell it where the Gurobi library is. I'm not sure how you do that in CLion, but for example if you were running CMake from the command line, you'd do:
cmake . -DCMAKE_PREFIX_PATH=/Library/gurobi604/mac64
I have two versions of boost:
The first one (from xcode i guess) has files in /usr/local/include/boost and
/usr/local/lib/boost
The second one (from homebrew) has files in
/usr/local/Cellar/boost/1.54.0/include/boost and
/usr/local/Cellar/boost/1.54.0/lib
As I want to use the second one, I call cmake to create my makefile and i compile my project this way:
cmake -DBOOST_ROOT=/usr/local/Cellar/boost/1.54.0/ .
make
And I get this error:
/usr/local/include/boost/functional/hash/extensions.hpp:54:17: error:
variable 'hash_value' declared as a template
std::size_t hash_value(std::list<T, A> const& v);
^
Clearly the boost version that is used is not the one I mentioned using BOOST_ROOT . How do I make sure that the version of boost that is used is the one in /usr/local/Cellar/boost/1.54.0/ ?
Thanks in advance
My guess is that you have another dependency that requires -I/usr/local/include/ in the flags, and that that -I flags is seen by the compiler before -I/usr/local/Cellar/boost/1.54.0/include.
Debug your compilation process with:
make VERBOSE=1
to see the compilation commands that are used.