I've noticed that CMake is installed with numerous modules such as the IAR compiler:
https://github.com/Kitware/CMake/blob/master/Modules/Compiler/IAR.cmake
In a previous question I asked how to load this particular module. The answer was just to add:
set(CMAKE_C_COMPILER iccarm.exe)
on my CMakeLists.txt.
Unfortunately I noticed, this is not enough because the IAR.cmake is never really used. With the following CMakeLists.txt:
cmake_minimum_required(VERSION 3.6)
set(CMAKE_SYSTEM_NAME Generic)
set(CMAKE_SYSTEM_PROCESSOR arm)
set(CMAKE_C_COMPILER iccarm.exe)
set(CMAKE_FIND_ROOT_PATH_MODE_PROGRAM NEVER)
set(CMAKE_FIND_ROOT_PATH_MODE_LIBRARY ONLY)
set(CMAKE_FIND_ROOT_PATH_MODE_INCLUDE ONLY)
set(CMAKE_FIND_ROOT_PATH_MODE_PACKAGE ONLY)
list(APPEND src src/main.c)
project("foo" C)
add_executable(foo ${src})
message(STATUS "IARARM_CMAKE_LOADED=${_IARARM_CMAKE_LOADED}")
message(STATUS "IAR CMAKE_C_COMPILE_OBJECT=${CMAKE_C_COMPILE_OBJECT}")
I get this output:
-- The C compiler identification is IAR
-- Check for working C compiler: C:/Program Files (x86)/IAR Systems/Embedded Workbench 8.0/arm/bin/iccarm.exe
-- Check for working C compiler: C:/Program Files (x86)/IAR Systems/Embedded Workbench 8.0/arm/bin/iccarm.exe -- works
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- IARARM_CMAKE_LOADED=
-- IAR CMAKE_C_COMPILE_OBJECT=<CMAKE_C_COMPILER> <SOURCE> <DEFINES> <INCLUDES> <FLAGS> -o <OBJECT>
-- Configuring done
-- Generating done
-- Build files have been written to: C:/Users/Ycr/Home/sandbox/cmake
Where IARARM_CMAKE_LOADED isn't defined and CMAKE_C_COMPILE_OBJECT doesn't have the --silent option defined here
How can I tell CMake to use the IAR Module?
I slowly begin to realize the issues.
you shall not define the compiler in the CMakeLists.txt file. you add them as argument on the commandline like cmake -DCMAKE_C_COMPILER=iccarm.exe ...
Compiler detection usually happens before your CMakeLists.txt file gets touched
Are you sure you are using the nightly build, because the file you linked is not in the relased builds but will be included in CMake 3.10
Add the versions and the commandline used to invoke CMake next time. Also you have some options like --trace-expand that should give you an idea what (does not) happens.
Related
Goal
I am using CMake to make the build for my project which is an embedded firmware based on ARM Cortex platform built using arm-none-eabi-gcc-6.3.1 compiler, using VSCode Editor, and on Windows host. I am trying to make a second build for testing on the Windows system I am using for the same project.
The First Solution Problem
The issue I am having is that whenever I need to switch my build from production to test, I have to delete the build files and rerun the CMake command with the test argument, when I do not do that, the build does not change the ARM compiler to the one I intend to use (I am guessing it is a caching problem). I have tried make clean and make rebuild_cache thinking that it may clean the cache and solve the problem for me, and did not work.
A fresh build Example:
Project\build> cmake -G"MinGW Makefiles" -DTARGET_GROUP=test ..
You have changed variables that require your cache to be deleted.
Configure will be re-run and you may have to reset some variables.
The following variables have changed:
CMAKE_C_COMPILER= C:/MinGW/bin/gcc.exe
CMAKE_CXX_COMPILER= C:/MinGW/bin/g++.exe
-- The C compiler identification is GNU 6.3.0
-- The CXX compiler identification is GNU 6.3.0
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: C:/MinGW/bin/gcc.exe - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: C:/MinGW/bin/g++.exe - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Configuring done
-- Generating done
-- Build files have been written to: Path\to\Project\build
Project\build>make
(successful make)
The second build:
Project\build> cmake -G"MinGW Makefiles" -DTARGET_GROUP=production ..
You have changed variables that require your cache to be deleted.
Configure will be re-run and you may have to reset some variables.
The following variables have changed:
CMAKE_C_COMPILER= Path/to/arm-gnu-toolchain/bin/arm-none-eabi-gcc.exe
CMAKE_CXX_COMPILER= Path/to/arm-gnu-toolchain/bin/arm-none-eabi-g++.exe
-- The C compiler identification is GNU 6.3.0
-- The CXX compiler identification is GNU 6.3.0
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: C:/MinGW/bin/gcc.exe - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: C:/MinGW/bin/g++.exe - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Configuring done
-- Generating done
-- Build files have been written to: Path\to\Project\build
Project\build>make
(Cached make -did not change the compiler or the files to be built)
The CMakeLists.txt:
set(TARGET_GROUP test CACHE STRING "Group to build")
if(TARGET_GROUP STREQUAL production)
# ARM Lib
include("arm-gnu.cmake")
else()
include("win-gcc-for-testing.cmake")
endif()
# include the files based on the TARGET_GROUP value
# ...
Current Solution
Currently, I have a temporary solution by making a separate folder for each building type and they work perfectly.
The question
Is the current solution (making two build directories each is for a different environment) right in terms of best practices? If not, what could be a better one?
As #Tsyvarev says, you have to use different directories if you want different builds. Here you are trying to build both in the same build directory. Instead, create a subdirectory test to build the -DTARGET_GROUP=test, and a different subdirectory prod to build the -DTARGET_GROUP=production:
Project> mkdir test
Project> cd test
Project\test> cmake -G"MinGW Makefiles" -DTARGET_GROUP=test ..
Project\test> cd ..
Project> mkdir prod
Project> cd prod
Project\prod> cmake -G"MinGW Makefiles" -DTARGET_GROUP=production ..
Of course you can call the directories whatever you like.
Since you'd like this to integrate into VSCode, do the configuration change with the CMake build kits, which is what sets the compiler prior to calling CMake. You can specify additional user kit definitions, and in the VSCode interface choose between compiler kits.
Rewrite the CMake files just enough to condition everything else on the selection of compiler.
THEN, the VSCode cmake-tools extension supports variable expansion for the name of the build directory. Among the options are ${buildKit} and ${buildKitTargetArch}. Make this suffix part of the Build Directory setting (your choice of whether to set this only in your workspace, or more globally on your system).
Now, when you switch kits, CMake will pick up the directory change and happily exist with two different build directories.
Alternatively, cmake-tools now provides variants, which could be used to add additional Production and Test to the normal Release, Debug, RelWithDebInfo, etc. There are additional build directory expansion variables for information from the currently selected variant.
I am trying to build a CMake project with MinGW. I have installed MinGW 8.1 through the Qt installer. The CMakeList is unchanged from how Qt Creator created it when creating the project.
When running CMake, I can see how Qt Creator passes the compiler and linker's paths to CMake and how CMake completely ignores them.
Führe C:\Program Files\CMake\bin\cmake.exe -S [snip] -B [snip] "-DCMAKE_BUILD_TYPE:String=Debug" "-DQT_QMAKE_EXECUTABLE:STRING=C:/Qt/5.15.1/mingw81_64/bin/qmake.exe" "-DCMAKE_PREFIX_PATH:STRING=C:/Qt/5.15.1/mingw81_64" "-DCMAKE_C_COMPILER:STRING=C:/Qt/Tools/mingw810_64/bin/gcc.exe" "-DCMAKE_CXX_COMPILER:STRING=C:/Qt/Tools/mingw810_64/bin/g++.exe" in [snip] aus.
-- Building for: Visual Studio 16 2019
-- Selecting Windows SDK version 10.0.18362.0 to target Windows 10.0.19041.
-- The CXX compiler identification is MSVC 19.24.28314.0
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: C:/Program Files (x86)/Microsoft Visual Studio/2019/Community/VC/Tools/MSVC/14.24.28314/bin/Hostx64/x64/cl.exe - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Configuring done
-- Generating done
CMake Warning:
Manually-specified variables were not used by the project:
CMAKE_BUILD_TYPE
CMAKE_C_COMPILER
QT_QMAKE_EXECUTABLE
Here I am using a system install of CMake, but the output is unchanged when using the CMake installed with Qt Creator. The CMake generator for the selected kit is set to MinGW Makefiles in the Qt Creator settings. Running CMake manually, without Qt Creator, allows me to configure the project. (although it still doesn't compile because of a linker error, that I expect to be independent of this problem)
It seems there is a caching issue. The caching is done by CMake (CMakeCache.txt) and by QtCreator (CMakeLists.txt.user) itself.
Maybe this helps:
Close QtCreator (by closing QtCreator the CMakeLists.txt.user file is saved or generated if the file don't exist)
delete QtCreator's CMakeLists.txt.user
the build folder
Open the project again and select the desired kit.
I would like to build a minimal example to build a C program. I used this CMakeLists.txt file:
cmake_minimum_required(VERSION 3.6)
set(CMAKE_SYSTEM_NAME Generic)
set(CMAKE_SYSTEM_PROCESSOR arm)
set(CMAKE_C_COMPILER "arm-none-eabi-gcc")
#set(CMAKE_CXX_COMPILER "arm-none-eabi-g++")
set(CMAKE_EXE_LINKER_FLAGS "--specs=nosys.specs" CACHE INTERNAL "")
set(CMAKE_FIND_ROOT_PATH_MODE_PROGRAM NEVER)
set(CMAKE_FIND_ROOT_PATH_MODE_LIBRARY ONLY)
set(CMAKE_FIND_ROOT_PATH_MODE_INCLUDE ONLY)
set(CMAKE_FIND_ROOT_PATH_MODE_PACKAGE ONLY)
project("funambule")
list(
APPEND src
main.c
)
add_executable(
funambule
${src}
)
When I run cmake .. CMake absolutely want to check the C++ compiler even though I don't need one. How can I prevent it to do this useless check?
-- The C compiler identification is GNU 5.4.1
-- The CXX compiler identification is GNU 5.4.0
-- Check for working C compiler: /cygdrive/c/Users/NoOne/Home/bin/arm-none-eabi-gcc
-- Check for working C compiler: /cygdrive/c/Users/NoOne/Home/bin/arm-none-eabi-gcc -- works
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - failed
-- Detecting C compile features
-- Detecting C compile features - done
-- Check for working CXX compiler: /usr/bin/CC
-- Check for working CXX compiler: /usr/bin/CC -- broken
...
This is the default behaviour of CMake. To change it, you should specify language for your project using:
project(<PROJECT-NAME> [LANGUAGES])
From CMake documentation:
Optionally you can specify which languages your project supports. Example languages are C, CXX (i.e. C++), Fortran, etc. By default C and CXX are enabled if no language options are given. Specify language NONE, or use the LANGUAGES keyword and list no languages, to skip enabling any languages.
in order to use colormake I did set this alias in my .bashrc
alias make="/usr/bin/colormake"
It works, as if I try to compile (with qmake) a simple C++ example code with errors (just a main.cpp with a cout ), they are correctly coloured.
However, if I compile the same code with cmake, colormake is not used. What can I do to force cmake to use it?
my minimal CMakeList.txt example is
CMAKE_MINIMUM_REQUIRED(VERSION 3.0.2)
ADD_DEFINITIONS(-std=c++11)
ADD_EXECUTABLE(exe main.cpp)
System: Debian 8.8 jessie
Thanks, Valerio
Update:
I modified the CMakeLists.txt in this way, but no success:
CMAKE_MINIMUM_REQUIRED(VERSION 3.0.2)
ADD_DEFINITIONS(-std=c++11)
set(CMAKE_MAKE_PROGRAM /usr/bin/colormake)
ADD_EXECUTABLE(exe main.cpp)
message("CMAKE_MAKE_PROGRAM: " ${CMAKE_MAKE_PROGRAM})
Update 2:
I modified the CMakeList in this way:
CMAKE_MINIMUM_REQUIRED(VERSION 3.0.2)
ADD_DEFINITIONS(-std=c++11)
#set(CMAKE_COLOR_MAKEFILE OFF)
#set(CMAKE_MAKE_PROGRAM /usr/bin/colormake)
ADD_EXECUTABLE(exe main.cpp)
message("CMAKE_MAKE_PROGRAM: " ${CMAKE_MAKE_PROGRAM})
message("CMAKE_COLOR_MAKEFILE: " ${CMAKE_COLOR_MAKEFILE})
then launched cmake with this argument from command line:
cmake -DCMAKE_MAKE_PROGRAM=/usr/bin/colormake -DCMAKE_COLOR_MAKEFILE=OFF ../
But again, the main.cpp synthax error after make is not coloured.
This is the output of cmake, note the messages about CMAKE_MAKE_PROGRAM and CMAKE_COLOR_MAKEFILE
-- The C compiler identification is GNU 4.9.2
-- The CXX compiler identification is GNU 4.9.2
-- Check for working C compiler: /usr/bin/cc
-- Check for working C compiler: /usr/bin/cc -- works
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working CXX compiler: /usr/bin/c++
-- Check for working CXX compiler: /usr/bin/c++ -- works
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
CMAKE_MAKE_PROGRAM: /usr/bin/colormake
CMAKE_COLOR_MAKEFILE: OFF
-- Configuring done
-- Generating done
-- Build files have been written to: /home/valeriosperati/Desktop/VALERIO_SPERATI/prova_codice_c/colormake/cmake/build
Some additional (maybe helpful) info: this is the output
I obtain when compiling with qmake, the error 'hjskf' is in red.
this is the output when comiling with cmake:
It should be enough to set the CMAKE_MAKE_PROGRAM cache variable to point at the build tool you want to use. You can do this by running cmake with a -D option like so:
cmake -DCMAKE_MAKE_PROGRAM=/usr/bin/colormake path/to/src
For the benefit of others, this technique can be used with other generators too, not just make. Just be aware that it is your responsibility to make sure the build tool specified matches the generator type CMake is using (i.e. don't pass a make tool if you've told CMake to use Ninja with -G Ninja instead).
Note, however, that this only really matters if you are invoking the build via CMake like so:
cmake --build path/to/build/dir
Some IDE tools may invoke the build that way. Most of the time, however, developers invoke the tool directly. In your case, you can simply invoke colormake instead of make. If you are still not getting colored output after doing that, then your problem must be elsewhere (check your terminal type settings perhaps).
Recently I've started using cmake instead of creating make-files manually. Moreover I use kdevelop as an IDE. So, I created simple cmake project with kdevelop. It builds and executes successfully. But the thing is that when I try to run cmake from terminal (without kdevelop involved in the process) I see that cmake just loads the cpu as high as possible and there is no result for about half an hour. I couldn't wait more so I've just kill the process.
Here is my cmake file:
project(robot)
cmake_minimum_required(VERSION 2.8)
set(CMAKE_BUILD_TYPE Debug)
include_directories(include)
add_library(mylib SHARED mylibsrc/mylib.cpp)
Here is how kdevelop starts runs cmake:
/home/sergey/projects/project-test/build> /usr/bin/cmake -DCMAKE_BUILD_TYPE=Debug /home/sergey/projects/project-test/
-- The C compiler identification is GNU 4.7.2
-- The CXX compiler identification is GNU 4.7.2
-- Check for working C compiler: /home/sergey/bin/gcc
-- Check for working C compiler: /home/sergey/bin/gcc -- works
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working CXX compiler: /home/sergey/bin/c++
-- Check for working CXX compiler: /home/sergey/bin/c++ -- works
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Configuring done
-- Generating done
-- Build files have been written to: /home/sergey/projects/project-test/build
I try to run cmake in same way but all I receive is the highest possible cpu load.
kdevelop version - 4.8.4
cmake version - 2.8.9
Can you advice anything about that?
Sorry for my broken English.
You can try adding the --trace option to the cmake call. The problem will still exist, but at least you should see then what is taking so long and can then further investigate. The --debug-output option might also help.
/usr/bin/cmake -DCMAKE_BUILD_TYPE=Debug --trace --debug-output /home/sergey/projects/project-test/