I've written a module on top of a private fork off of TensorFlow that uses nanomsg.
For my local development server, I used cmake install to install nanomsg (to /usr/local) and accessed the header files from their installed location. The project runs fine locally.
However, I now need to package nanomsg within my TensorFlow workspace. I've tried the following two approaches, and find neither satisfactory:
Similar to this answer for OpenCV, I precompiled nanomsg into a private repository, loaded it within my workspace (within tensorflow/workspace.bzl) using an http_archive directive then included the headers and libraries in the relevant build script. This runs fine, but is not a portable solution.
A more portable solution, I created a genrule to run a specific sequence of cmake commands that can be used to build nanomsg. This approach is neater, but the genrule cannot be reused to cmake other projects. (I referred to this discussion).
Clearly cmake is not supported as a first-class citizen in Bazel builds. Is there anyone who has faced this problem in your own projects created a generic, portable way to include libraries within Bazel projects that are built using cmake? If so, how did you approach it?
As Ulf wrote, I think your suggested option 2 should work fine.
Regarding "can I identify if the cmake fails", yes: cmake should return with an error exit code (!= 0) when it fails. This in turn will cause Bazel to automatically recognize the genrule action as failed and thus fail the build. Because Bazel sets "set -e -o pipefail" before running your command (cf. https://docs.bazel.build/versions/master/be/general.html#genrule-environment), it should also work if you chain multiple cmake commands in your genrule "cmd".
If you call out to a shell script in your "cmd" attribute that then actually runs the cmake commands, make sure to put "set -e -o pipefail" in the first line of your script yourself. Otherwise the script will not fail when cmake fails.
If I misunderstood your question "Can I identify if the cmake fails", please let me know. :)
This new project: https://github.com/bazelbuild/rules_foreign_cc seems like a solution(it build rules for cmake to build your project inside bazel).
Related
I'm trying to get cmake working on msys2. I try to compile some code that I've been given for another project and:
CMake Error: Could not create named generator MSYS Makefiles
I've found this thread and this thread and tried to follow the instructions...only to realize that the wiki that the threads eventually link back to never states where msys stores toolchains. Googling around didn't really help.
Even more oddly, the program that I'm trying to use somehow runs and produces valid data despite throwing an error due to the missing cmake. I don't get it.
I've never used msys before; as someone who's brand new to msys2/cygwin and can't even understand the lingo, how do I get cmake installed and working?
Sounds like you are specifying a -G "MSYS Makefiles" as the CMake generator, but CMake doesn't recognize that generator. When I use CMake on msys2 I just use the default Unix Makefiles generator and everything works just fine. Also, on my current msys2 install, CMake doesn't seem to have a MSYS Makefile generator that I can see (running cmake --help will list the available generators). Try running cmake without the -G option. Also, make sure make is installed first via pacman -Sy make.
I'm trying to use CppUTest in Windows, first step is to get it to work and I already have problems. These are the things I've tried:
First Approach
With CMake, using the cmake GUI I can do the configure and generate command and I get something in the output directory, but no binaries and no libraries, just a bunch of cmakefiles. The CMake GUI says everything went OK during the configuration and generation steps, however the libraries (.lib files) are not generated in the output directory... is there something I am missing? I've never used CMake before.
Second approach
With MinGW and msys alone, running cmd in Windows and executing a MinGW shell by typing sh in the Windows terminal, afterwards I execute the following commands:
cd <CppUTest folder>
mount c:\mingw /mingw
./autogen.sh
./configure
make
The build process starts but it fails with a message indicating that pthread.h was not found in MinGW directory. If I install the pthread-win32 package with the MinGW package manager and repeat the same steps as above the build process starts but fails with a message indicating that the structure timespec is defined in time.h and pthread.h.
I've tried to follow this same procedure with CppUTest 3.6 and it works perfectly fine, I get the .lib files, so I guess I will have to continue with this for now.
Does anyone know how to build CppUTest 3.7 (latest release) with MinGW or CMake?
In the end I used Cygwin to compile it, I couldn't find a way to compile it with MinGW properly, I added a dirty trick to make it compile under MinGW (handled the timespec redifinition) but chances are that is going to cause issues.
Just make sure that you use Cygwin aswell to compile your tests, something that I found out after making this question (https://www.youtube.com/watch?v=oVmd0P85D8o).
When building a CMake project (e.g. on Windows), I can perform a parallel build by executing the following command:
cmake --build . -- /m
The /m switch is passed to msbuild and I get parallelized builds. However, if I have some external project in my CMakeLists, that switch is not passed through to its build command.
What's the best way to ensure my ExternalProject is built in parallel? I don't want to hardcode a BUILD_COMMAND if I can avoid it.
In CMake 3.12, support was added to kick off parallel jobs from within cmake.
The previous versions only supported this via passthrough arguments after the --, which understandably wouldn't pass through to external builds. Though, I'm not sure if the new support works for ExternalProject, it'd be great to try and report back if it works!
I had the same problem with make based external_projects. I ended up with the solution to pass the -j8 flag via a cmake cached variable.
set(EXTERNAL_BUILD_ARGS "/m" CACHE STRING "flags for external project builds")
My cmake --version is 2.8.12.2.
I configure my project build with these commands:
cmake ../klein/ -DBUILD_KLEIN_DEPS=1 -DCMAKE_EXPORT_COMPILE_COMMANDS=1
cmake ../klein/ -DBUILD_KLEIN_DEPS=1 -DCMAKE_EXPORT_COMPILE_COMMANDS=ON
cmake ../klein/ -DBUILD_KLEIN_DEPS=1 -DCMAKE_EXPORT_COMPILE_COMMANDS=on
CMAKE_EXPORT_COMPILE_COMMANDS=1 cmake ../klein/ -DBUILD_KLEIN_DEPS=1
From a clean build, and from a directory with an existing successful build. And want to see the compiler_commands.json file, but it does not appear.
At which moment should it be created: after cmake, or after make command? Where should it be in ./, or in ../klein directory? My cmake does not say anything about this option while it always complains about unused build variables.
Should it work from in a "dirty" directory, where I've performed one successful build, or does it work only on a fresh run in an empty folder?
Edit:
I use a default generator "Unix Makefiles" on my ubuntu linux machine
Edit2:
I'm not an author of the project under the question (I just want to explore it with rtags which requires compile_commands.json file), and I'm not very familiar with CMake mechanics. However, the CMakeLists.txt is probably configured as a super-build (it indeed downloads and builds dependencies - like llvm, z3, ...), and it includes ExternalProject, however it also builds the project itself (klein) from sources. So it's a mix, as I would say.
Can you specify what generator your using? A quick scan of the cmake source from version 3.1.0 suggests that this command is still only available in the following 2 cases.
if(CMAKE_GENERATOR MATCHES "Unix Makefiles")
and
if(CMAKE_GENERATOR MATCHES "Ninja")
if you're using Visual Studio directly you're out of luck unless you want to add a patch to CMake. Otherwise, I know many Windows developers who've gone to Ninja. One advantage is that it's vastly faster than Visual Studio for building. If you are, in fact using Ninja or Unix Makefiles, then it's worth digging deeper.
I'm new to cmake and I'm finding it very frustrating. I am trying to use wildcards in file paths that are evaluated when the build runs, not when the build is generated.
I have created a build that uses SWIG to generate Java wrappers for some C++ code. I can write the commands to generate the native code, compile it, and produce a working shared library, and even use the INSTALL command to install that shared library correctly. What I can't figure out how to do is to write an INSTALL command that can copy all *.java files generated by SWIG into that same install location.
It seems that cmake's FILE GLOB command does the globbing when cmake is executed, and not when the build actually runs. Of course, when cmake is executed, SWIG hasn't run yet, and the Java files don't exist.
Is there a way to do what I want? Am I going about things wrong? It seems like this is such a fundamental part of what Makefiles need to do, I'm really surprised not to find an easy way to do it.
Assuming that the Java wrappers are located in the current binary directory, you can use the following install command to copy the Java files upon install:
install(
CODE "file( GLOB _GeneratedJavaSources \"${CMAKE_CURRENT_BINARY_DIR}/*.java\" )"
CODE "file( INSTALL \${_GeneratedJavaSources} DESTINATION \"$ENV{HOME}\" )"
)
The CODE form of the install command is used to execute two CMake commands upon running the install target. The first one collects all generated Java files in a helper variable. The second one uses the INSTALL form of the file command to copy the files.
you can use install(SCRIPT swigInstaller.cmake) or install(DIRECTORY) both of which supports doing file globing at install time. You can read more about the install command at:
http://cmake.org/cmake/help/cmake-2-8-docs.html#command:install