We have a quite large Fortran program that's developed and builded using Visual Studio and Ifort for Windows. On the linux side, we used to get the .dep generated on windows side, proccess it using a script, and create a command line call with the files on the order needed for the compilation. Now, I'm trying to rewrite the linux build using CMake.
After many solved problems (casing, enconding, ...) I've encountered some errors on the linking part. We have some subroutines defined on separated files that get called straight from other files like: CALL SUBROUTINE_NAME ()
I guess VisualStudio is somehow handling this linking but when I try to compile it using CMake on linux i get an error "undefined reference to SUBROUTINE_NAME".
My first tought was trying to include the file where this routine is defined but the subroutines are defined on multiple directories and, as far I understand, unless I put all this places on the include path for the compiler they can't be found.
Another option would be rewriting all this as modules and using them where needed. I guess this would be a cleaner solution but also a lot of work that I'm not sure I'd have the time for now.
First of all, I'd like to make sure my reasoning seems right and then what would be the best or some alternatives way to deal with this, like a way to manually include the paths for each occurrence or if there's a way to make CMake work like VisualStudio and resolve this references for example.
Thanks in advance for any insight and don't hesitate to get in touch and ask for more information if needed.
Edit to add more CMake info as requested:
CMakeLists.txt:
file(GLOB_RECURSE sources ./src/*.for)
add_executable(program_name ${sources}
target_include_directories(program_name PUBLIC "./src/includes")
I've managed to manually compile the file passing the .o of the separated subroutine but couldn't get CMake to include it as needed.
My subroutine file is been processed by CMake and a .o file is been generated for it, as I'm passing it as source.
Whenever I use a module, CMake can automatically add the module as a dependency for the file. But on this cases, where I'm calling the subroutine directly it doesn't generate the correct dependency for the Makefile.
Related
I have an embedded project (using ESP-IDF which builds projects with CMake), where I have a props.json file that contains several settings (e.g. "device type"). For example based on the actual value of "deviceType" the CMake open and read props.json by calling execute_process() and jq, then defines C preprocessor macros, such as: DEVICE_TYPE_A by using add_compile_definitions().
The problem is that, this will run only when I modify the CMakeLists.txt or clean the whole project, but I don't want to recompile each components when I change the props.json only the files that I wrote (so, depend on the settings). I'd like to make CMake read the file each time I build the project without cleaning it.
I did my research, so I know there are add_custom_target() and add_custom_command() that behave that way, however add_compile_definitions() cannot be called in a script. Is there a solution to achieve this or should I just use a header file configured by configure_file() and leave add_compile_definitions() alone?
This is actually pretty easy and you don't need to manually reconfigure CMake. Just add the following to the CMakeLists.txt in the directory containing your props.json file:
set_property(DIRECTORY . APPEND PROPERTY CMAKE_CONFIGURE_DEPENDS props.json)
This will add props.json to the list of files that the CMake-generated build scans when determining whether to re-run the CMake configure step. See the docs on CMAKE_CONFIGURE_DEPENDS for more detail.
In general, you should never need to manually re-run CMake1 after the first configure. If you do, it is an indication that you have not communicated all of the necessary information for CMake to generate a correct build system.
1 There is one notable exception: Xcode is known to be buggy when re-running the CMake configure step automatically.
I currently want to create an installer with cmake, but don't add all necessary DLLs by myself to CMakeLists.txt. So one solution should be to use fixup_bundle, like here suggested, so hopefully he copy all DLLs, which he can detect with a dependency walker and are on path.
But currently I have no idea how is best way to use it on a target, following code won't work, because he don't resolve TARGET_FILE_DIR like if you are using add_custom_command. Do read location via get_property won't work too, because he don't know the target anymore at time of execution. Any idea?
INSTALL(CODE "
include(BundleUtilities)
fixup_bundle($<TARGET_FILE_DIR:${PROJECT_NAME}> \"\" \"D:\\Qt\")
" COMPONENT Runtime
)
If you are using Qt4, rather than using BundleUtilities directly, you may be better off using the DeployQt4 module. It includes the following three commands which may do what you need:
install_qt4_plugin_path
install_qt4_plugin
install_qt4_executable
If you are using Qt5, it gets a bit trickier. If you are only interested in Windows and/or Mac, then Qt itself provides an appropriate tool for handling/bringing across Qt's dependencies. The relevant tools are called windeployqt and macdeployqt respectively. Sadly, at time of writing, there is no linuxdeployqt tool yet that I'm aware of.
If neither of the above options are open to you or don't do what you need, then at least the DeployQt4 module gives some clues as to how you may be able to use the INSTALL(...) command like you attempted to. The DeployQt4 module uses the following for defining its target (see right near the end of the DeployQt4.cmake file):
FIXUP_QT4_EXECUTABLE(\"\$ENV{DESTDIR}\${CMAKE_INSTALL_PREFIX}/${executable}\" \"\" \"${libs}\" \"${dirs}\" \"${plugins_dir}\" \"${request_qt_conf}\")"
${component}
)
The stuff in front of ${executable} is probably the bit you were missing. In your case, without seeing your full CMakeLists.txt file, I can only assume that you have a single target and it has the same name as your project (since you used ${PROJECT_NAME} in your attempted generator expression). You could try something like the following (not tested):
INSTALL(CODE "
include(BundleUtilities)
fixup_bundle(\"\$ENV{DESTDIR}\${CMAKE_INSTALL_PREFIX}/${MyTarget}\" \"\" \"D:\\Qt\")
" COMPONENT Runtime
)
where MyTarget is the name of the executable for your target (I think without any .exe suffix if you are on Windows). The DESTDIR part is needed when making packages, since CMake will redirect the install location by setting the DESTDIR environment variable (at least with some CMake generators). The CMAKE_INSTALL_PREFIX part is the path under which the application would be installed. There is some history behind this, but the above reflects the correct way of how to refer to the installed executable.
I am using a 3rd party library rbdl, which contains rbdl.pc.cmake, which 'I suppose' is included for using pkg_check_modules in a cmake file.
I update PKG_CONFIG_PATH to point at the rbdl folder
set(ENV{PKG_CONFIG_PATH} "$ENV{PKG_CONFIG_PATH}:${CMAKE_BINARY_DIR}/externals/rbdl")
pkg_check_modules(RBDL rbdl)
but pkg_check_modules says it cannot find the module.
When I manually duplicate rbdl.pc.cmake, rename the copy into rbdl.pc and run pkg-config --cflags --libs rbdl in terminal, then my cmake also start working!
Interestingly, now even if I delete rbdl.pc, rbdl module if perfectly found by rbdl.pc.cmake!
So my questions are:
What is the difference between *.pc and *.pc.cmake?
How do I correctly setup my cmake to work with original rbdl.pc.cmake?
Why rbdl.pc.cmake starts to be accepted by pkg_check_modules after that tweak with duplicating it, renaming the copy and running pkg-config manually?
You understand it wrong! rbdl.pc.cmake is just a template file. It is not supposed to be used by you! Take a look into CMakeLists.txt line 160 -- configure_file() used to render variables ("quoted" by # in template file) and produce a rbdl.pc (a real pkg-config) file. Latter should be installed (some way) and then will be available to pkg-config hence can be used in your project.
pkg-config is stupid do not interpret or validate compiler/linker flags any way, so your renamed file "works" (yeah, producing invalid command line for compiler/linker).
I wish you to read CMake documentation before trying to code something using it! It'll save your time and give you a necessary knowledge which stops you from doing stupid things like you described in your question ;-)
I was trying to parallelize a Fortran program using MPI. I use CMake to do the build of my program. It was difficult to find support on getting CMake to create a working makefile for Fortran with MPI support on google, but from what I could gather, I added the following commands to my CMakeLists.txt script:
find_package(MPI REQUIRED)
add_definitions(${MPI_Fortran_COMPILE_FLAGS})
include_directories(${MPI_Fortran_INCLUDE_DIRS})
link_directories(${MPI_FortranLIBRARY_DIRS})
This will locate MPI on my system and then set the variables found in the following three commands. In my linking line, I added the MPI libraries variable to the list of the other libraries that my program needed to build.
target_link_libraries(${exe_name} otherlibs ${MPI_FortranLIBRARY_DIRS})
Doing cmake and then make worked to build the program and the program ran; however, when I tried to add more to the source which required me to include the mpif.h include file, my compilation failed due to not being able to find this header file. I also could not use mpi because the compiler cannot find the mpi.mod file in the path.
I inserted "message" commands into my CMakeLists.txt file and printed out the values of the variables that I was using for including and linking. It turns out that the variables, MPI_Fortran_INCLUDE_DIRS and MPI_FortranLIBRARY_DIRS weren't set to anything. A check of the module that CMake is actually using to set these variables (FindMPI.cmake) showed these variables to be non-existent. I changed my CMakeLists.txt file to use the correct variables:
find_package(MPI REQUIRED)
add_definitions(${MPI_Fortran_COMPILE_FLAGS})
include_directories(${MPI_Fortran_INCLUDE_PATH})
link_directories(${MPI_Fortran_LIBRARIES})
target_link_libraries(${exe_name} otherlibs ${MPI_Fortran_LIBRARIES})
Now when I execute make, the compiler could find both mpif.h as well as mpi.mod.
UPDATE:
I want to mention that this solution worked for cmake version 2.8.10.1. When I moved my CMakeLists.txt scripts to a different machine that has cmake version 2.8.4, I get the same error about mpi.mod missing during the link stage. I checked the FindMPI.cmake module and, sure enough, there are no variables that specify the language (i.e. there is no MPI_Fortran_LIBRARIES variable, just a MPI_LIBRARIES variable, and this variable is not getting set to the correct location of the mpi library on that system. So this solution will be dependent on cmake version.
Sounds like you are not using the mpi compiler. That is fine, but you have to add a lot of flags then. There is not really an mpi compiler but a wrapper that sets the flags to be able to use mpi. With cmake I was able to do this by defining the fortran compiler I was going to use BEFORE the call to cmake. It's not a nice solution since you loose portability, but it works. I'm trying to find a better solution and define inside cmake what compiler to use, but have not been able to do so. In other words, this works for me:
FC=mpif90 cmake ../.
make
I was having the same problem as you. Hope this solves the issue. If anybody finds how to define the fortran compiler INSIDE cmake please post it!
as you've already noticed, you misspelled the name of two variables, namely MPI_Fortran_LIBRARIES and MPI_Fortran_LIBRARIES
It is useful also to add:
cmake_minimum_required(VERSION 2.8.10)
at the very beginning of your CMake to be sure that these variables will be defined.
I have a specific question which serves as context for a more general question.
There is a scientific package called LAMMPS, and it is usually used as an executable. However, it supports use as a "library". To try to do things right, I put it in /usr/local/lib/lammps. It contains a lammps/src/ directory, which has around 40 source files. Using the instructions provided, I compiled lammps as a .so file in lammps/src/liblammps_serial.so.
I also have separate code in "~/code/ljtube/". This uses cmake to try to find the library. Thus, I wrote a FindLAMMPS.txt so that I could use
FIND_PACKAGE (lammps)
in my CMakeLists. I modified the libtool config file to search in /usr/local/ successfully. I found that it searches in /usr/local/lib/ for a .so file and in /usr/local/include/ for a .h file. So I made a dynamic link to the .so file in /usr/local/lib/, and I copied the .h file from the lammps/src/ to /usr/local/include/.
CMake can now find those two files, but it cannot link to anything else in lammps/src/. It seems absurd to need to make a separate FIND_PACKAGE for each of the .h's I want to include (group.h, fix.h, force.h, pair.h, etc.). It also seems ridiculous to dump the whole package of .h files into the /usr/local/include/ directory. I will be using this code both locally and on a cluster, and possibly distributing it to other group members.
How can I make CMake find what I want to find without hard coding in the location of /usr/local/lib/lammps/src/? Phrased more generically, how should I manage large packages like these to make them easy to link to in the code I write, even if the original developer did not use the best conventions?
(As a side note, I am using a shared library because it seems like the right choice, but I'm not especially married to it. Should I be using a static library? Is there a way for CMake to find an already-compiled library relative to the current source directory, and might that be a better way to implement this? I know that I will be using LAMMPS in multiple projects, so having a local shared copy superficially seems to make the most sense.)
Normally a find_package call yields a variable specifying the path to the "includes" folder of the package. This would then be added in the caller's CMakeLists.txt via include_directories.
For example, to use find_package for boost, you could do:
find_package(Boost) # sets ${Boost_INCLUDE_DIRS} and ${Boost_LIBRARIES}
if(Boost_FOUND)
include_directories(${Boost_INCLUDE_DIRS})
add_executable(foo foo.cc)
target_link_libraries(foo ${Boost_LIBRARIES})
endif()
Regarding your side note, you could use find_library and/or find_path to find the library and its headers given a known location.
Both these commands can be invoked in such a way as to avoid searching in common locations, e.g. by setting PATHS to the known location and using NO_DEFAULT_PATH in the find commands.
Another alternative is for your projects to make use of the ExternalProject_Add function which is described in more detail in this article. From this article:
The ExternalProject_Add function makes it possible to say “download this project from the internet, run its configure step, build it and install it”
A downside to this approach is that each of your projects would end up with its own copy of the third party sources and lib.