I have the following configuration in the .pro file
TEMPLATE = app
CONFIG += console c++11
CONFIG -= app_bundle
CONFIG -= qt
CONFIG += thread
SOURCES += main.cpp
INCLUDEPATH += /usr/local/include/opencv4
LIBS += -L/usr/local/lib/
LIBS += -lopencv_core
LIBS += -lopencv_highgui
LIBS += -lopencv_imgproc
LIBS += -lopencv_videoio
QMAKE_CXXFLAGS += -D_GLIBCXX_USE_CXX11_ABI=0
INCLUDEPATH += /path/to/libtorch/include
INCLUDEPATH += /path/to/libtorch/include/torch/csrc/api/include
LIBS += -L/path/to/libtorch/lib
LIBS += -ltorch -lc10
OpenCV works absolutely fine without "QMAKE_CXXFLAGS += -D_GLIBCXX_USE_CXX11_ABI=0". With this, however, I get this following errors:
OpenCV works fine with "QMAKE_CXXFLAGS += -D_GLIBCXX_USE_CXX11_ABI=1" as well. But it throws a different set of errors:
Setting "QMAKE_CXXFLAGS += -D_GLIBCXX_USE_CXX11_ABI=0" has been recommended for Libtorch in most of the forums to avoid the errors above.
What could be a solution or some solutions to work around this?
(I am a newbie to both Libtorch and Qt Creator.)
Maybe OpenCV and Libtorch were compiled with a different version of GCC (and different values of _GLIBCXX_USE_CXX11_ABI).
Try recompiling them by yourself and see if things change.
The problem is that you downloaded the wrong ABI version of LibTorch. It looks like that you downloaded the Pre-CXX11 ABI version of LibTorch and OpenCV is compiled with CXX11 ABI. So if you set _GLIBCXX_USE_CXX11_ABI=0, OpenCV throws errors, and if you set _GLIBCXX_USE_CXX11_ABI=1, LibTorch throws errors.
Download the CXX11 ABI LibTorch from PyTorch official website and you won't need to set the _GLIBCXX_USE_CXX11_ABI flag.
About the dual ABI problem of GCC, see https://gcc.gnu.org/onlinedocs/libstdc++/manual/using_dual_abi.html
Related
I tried to link against Vulkan with:
LIBS += -L$$(VK_SDK_PATH)/Bin -lvulkan-1 # Doesn't work at all
LIBS += -L$$(VK_SDK_PATH)/Lib -lvulkan-1 # Works with MSVC
But I couldn't get it linking with MinGW.
INSTALLED:
VulkanSDK (1.2.148.1)
Qt (5.15.0; MSYS2)
MinGW (10.2.0; MSYS2)
I installed the Vulkan Runtime and retried
LIBS += -L$$(VK_SDK_PATH)/Lib -lvulkan-1
with MinGW again and it works. Nevertheless, thank you for helping me out.
I am reading the LLVM's Kaleidoscope tutorial (http://llvm.org/docs/tutorial/index.html).
I wanted to compile and test the language. After some compiler's errors (EngineBuilder and Module's constructor, linking libs...), the example program was built. Then, I tried the language. I got a few problems with InitializeNativeTargets, DataLayoutPass... But I managed to correct them.
Howewer, I don't manage to resolve one error. When I write extern printd(x); printd(5);, the program doesn't work : "LLVM ERROR : Program used external function 'printd' which could not be resolved".
I looked for the solution on the net. I read a lot of webpages, but nothing worked.
How can I resolve this problem ? Why LLVM can't find the external function 'printd', which is included in the program ? Thanks in advance for your answer.
The used code : https://docs.google.com/document/d/1Qb-zUGaUUIF354uFCXv1iuq8n_rjya6IHDW4WCPWN_4/edit?usp=sharing
The .pro file (Qt Creator) :
QT += core
QT -= gui
TARGET = Kaleidoscope
CONFIG += console
CONFIG -= app_bundle
TEMPLATE = app
SOURCES += main.cpp
INCLUDEPATH += /home/alainetflo/Documents/me/C++/LLVM/llvm-3.5.0.src/include
DEPENDPATH += /home/alainetflo/Documents/me/C++/LLVM/llvm-3.5.0.src/include
LIBS += `llvm-config --ldflags --libs all --system-libs` -Wl,-no-as-needed
QMAKE_CFLAGS += -m32
QMAKE_CXXFLAGS += -rdynamic -std=c++11 -O3
I use GCC 4.8.2, LLVM 3.5 and Qt 5.3.1 on Ubuntu 14.04 32bits.
Finally, I found a great link on the net : http://koichitamura.blogspot.fr/2011/01/since-i-went-to-held-several-weeks-ago.html.
The error came from the wrong place of the -rdynamic argument in the .pro file (-rdynamic must be after the linking options). You need this argument because (GCC man page) :
This instructs the linker to add all symbols, not only used ones, to the dynamic symbol table.
I changed the .pro file :
QT += core
QT -= gui
TARGET = Test01
CONFIG += console
CONFIG -= app_bundle
TEMPLATE = app
SOURCES += main.cpp
INCLUDEPATH += /home/alainetflo/Documents/me/C++/LLVM/llvm-3.5.0.src/include
DEPENDPATH += /home/alainetflo/Documents/me/C++/LLVM/llvm-3.5.0.src/include
INCLUDEPATH += /usr/include/i386-linux-gnu/c++/4.8 \
/usr/include/c++/4.8
LIBS += `llvm-config --ldflags --libs all --system-libs` -Wl,-no-as-needed -rdynamic
QMAKE_CFLAGS += -m32
QMAKE_CXXFLAGS += -g -std=c++11 -O3
In this way, the program works with no error ! (The tutorial's example runs (http://llvm.org/docs/tutorial/LangImpl6.html#kicking-the-tires))
I'm trying to use the Sha512 function in openSSL but can't seem to get it to work as I get compiler errors just starting into the code. I include #include <openssl/sha.h> at the top of the .cpp file, then in the action of a button event I put just the following code below.
SHA512_CTX ctx;
SHA512_Init(&ctx);
//Will uncomment below in later if I get SHA512_Init to work
//SHA512_Update(&ctx, string, strlen(string));
//SHA512_Final(digest, &ctx);
I get a linker error telling my undefined symbols for architecture x86_64, implying the function does not exist?
I'm aware QT 5 has a hash function, but I'm limited to QT 4.8 so I can not use the cryptographic sha512 hash function available in the QT 5+ framework.
Any help is appreciated!
Used macports to install openssl
I'm using Mac OS 10.9.2
MAKE FILE
#-------------------------------------------------
#
# Project created by QtCreator 2014-06-11T20:27:49
#
#-------------------------------------------------
QT += core gui network
greaterThan(QT_MAJOR_VERSION, 4): QT += widgets
TARGET = ExchangeTab
TEMPLATE = app
LIBS += -L/usr/include/openssl -openssl
INCLUDEPATH += /usr/include/openssl
SOURCES += main.cpp\
mainwindow.cpp \
httpsocket.cpp \
cloaksend.cpp \
exchange.cpp
HEADERS += mainwindow.h \
httpsocket.h \
cloaksend.h \
exchange.h
FORMS += mainwindow.ui
RESOURCES += \
AppResources.qrc
Looking for cross platform solution please.
You need to link to the right library (openSSL)
Have a look here: How to Include OpenSSL in a Qt project
Specifically, add this to your .pro file.:
LIBS += -L/opt/local/lib/ -lcrypto
For including .h files add this line to your .pro file :
INCLUDEPATH += /opt/local/include
[1] says that the default include path will be /opt/local/include/.
LIBS += -L/usr/include/openssl -openssl
INCLUDEPATH += /usr/include/openssl
This looks incorrect. The OpenSSL libraries are libcrypto (-lcrypto) and libssl (-lssl). There is nolibopenssl(-lopenssl). Try:
LIBS += -L/usr/lib -lcrypto
INCLUDEPATH += /usr/include/openssl
But the libraries are version 0.9.8. You might consider upgrading to 1.0.1h.
$ ls /usr/lib | grep crypto
libcrypto.0.9.7.dylib
libcrypto.0.9.8.dylib
libcrypto.dylib
libk5crypto.dylib
And
$ /usr/bin/openssl version
OpenSSL 0.9.8y 5 Feb 2013
If you choose to upgrade, OpenSSL will install into /usr/local/ssl. Avoid mixing/matching version of OpenSSL with the following.
INCLUDEPATH += /usr/local/ssl/include/openssl
LIBS += /usr/local/ssl/lib/libcrypto.a
Its OK to specify objects and archives in LIBS. See How to add object files to a project in Qt.
Okay so I may have answered my own question but I will need some help understanding why it worked.
My make file was indeed the problem. I added the following 3 lines.
INCLUDEPATH += $$OPENSSL_INCLUDE_PATH
LIBS += $$join(OPENSSL_LIB_PATH,,-L,)
LIBS += -lcrypto
Then it magically compiled just fine. I found these in another project made with QT that compiled for OpenSSL.
Interestingly enough. I removed the top two lines so only the following remained. Then ran the clean on my project to be sure the code was being recompiled.
LIBS += -lcrypto
This also just 'worked' without linker errors. It looks like that is the only command I need. Question is... will that be cross platform friendly if I take this code and compile on linux or windows? That I am not sure, but this worked.
I have compiled TBB from source using Mingw following the comment #5 in this post: http://software.intel.com/en-us/forums/topic/291331. That went ok.
When I try to use the new TBB library in a QtCreator project, I end with this errors (ignore the warning messages): http://postimage.org/image/yrrecugix/
Here's the sample code I tried (I omit the non-tbb code):
#include "tbb/task_scheduler_init.h"
int main()
{
tbb::task_scheduler_init init;
/// more things.
}
And here's the .pro file:
TEMPLATE = app
CONFIG += console
CONFIG -= qt
SOURCES += main.cpp
#QMAKE_CXXFLAGS += -fopenmp
#QMAKE_LFLAGS += -fopenmp
INCLUDEPATH += "E:\TRABAJO\LIBRERIAS\tbb-4.1_src\include"
LIBS += -L"E:\TRABAJO\LIBRERIAS\tbb-4.1_src\build\windows_intel64_gcc_mingw4.5.4_debug\" \
-ltbb_debug
Any idea?.
Thanks!.
When built with MinGW on Windows, TBB binaries are tbb.dll and tbb_debug.dll. The option -ltbb_debug in your configuration files probably causes the linker to look for libtbb_debug.<something>. It can't find such a binary and reports about unresolved symbols.
Some days ago I compiled Boost ver. 1.53.0 for VS2012. It works fine, compiles fine. Now I want to use Boost with Qt Creator. In the .pro file I've included
INCLUDEPATH += C:\boost\boost_1_53_0\ -lboost_filesystem
LIBS += C:/boost/boost_1_53_0/stage/lib/
But when I compile I get 2 errors:
:-1: error: cannot find C:/boost/boost_1_53_0/stage/lib/: Permission denied
collect2.exe:-1: error: error: ld returned 1 exit status
What should I do? I've googled but seems I'm the first with this error.
INCLUDEPATH += C:\boost\boost_1_53_0\ -lboost_filesystem
LIBS += C:/boost/boost_1_53_0/stage/lib/
Wrong.
Read this.
Solution:
INCLUDEPATH += C:/boost/boost_1_53_0/
LIBS += "-LC:/boost/boost_1_53_0/stage/lib/"
Boost has complicated library names ("libboost_filesystem-vc90-mt-1_53.lib") and in case of msvc it links them automatically.)
If you want to link additional lib, you do it like this:
LIBS += "-LMyLibraryPath" -lmylib
Where MyLibraryPath is library path, and mylib is library you want to link with.
i'm the first with this error.
The error most likely occurs because compiler tries to open directory as if it were a file or something like that.
win32 {
INCLUDEPATH += C:/Users/User/Downloads/dev/boost_1_61_0
LIBS += "-LC:/dev/Boost/lib/" \
"-Llibboost_filesystem-mgw53-mt-d-1_61.a", "-Llibboost_system-mgw53-mt-d-1_61.a", "-Llibboost_serialization-mgw53-mt-d-1_61.a" -LLIBS
}