MesonToolchain
Warning
This is an experimental feature subject to breaking changes in future releases.
Available since: 1.33.0
The MesonToolchain
can be used in the generate()
method:
from conan import ConanFile
from conan.tools.meson import MesonToolchain
class App(ConanFile):
settings = "os", "arch", "compiler", "build_type"
requires = "hello/0.1"
options = {"shared": [True, False]}
default_options = {"shared": False}
def generate(self):
tc = MesonToolchain(self)
tc.preprocessor_definitions["MYDEFINE"] = "MYDEF_VALUE"
tc.generate()
The MesonToolchain
will generate a file:
- conan_meson_native.ini: if doing a native build.
- conan_meson_cross.ini: if doing a cross-build (tools.cross_building()).
Important
This class will require very soon to define both the “host” and “build” profiles. It is very recommended to start defining both profiles immediately to avoid future breaking. Furthermore, some features, like trying to cross-compile might not work at all if the “build” profile is not provided.
conan_meson_native.ini
will contain the definitions of all the Meson properties
related to the Conan options and settings for the current package, platform,
etc. This includes but is not limited to the following:
Detection of
default_library
from Conan settingsBased on existance/value of a option named
shared
Detection of
buildtype
from Conan settingsDefinition of the C++ standard as necessary
The Visual Studio runtime (
b_vscrt
), obtained from Conan input settings
conan_meson_cross.ini contains the same information as conan_meson_native.ini, but with additional information to describe host, target, and build machines (such as the processor architecture).
Check out the meson documentation for more details on native and cross files:
constructor
def __init__(self, conanfile, backend=None):
Most of the arguments are optional and will be deduced from the current settings
, and not
necessary to define them.
conanfile
: the current recipe object. Always useself
.backend
: the meson backend to use. By default,ninja
is used. Possible values: ninja, vs, vs2010, vs2015, vs2017, vs2019, xcode.
definitions
This attribute allows defining Meson project options:
def generate(self):
tc = MesonToolchain(self)
tc.definitions["MYVAR"] = "MyValue"
tc.generate()
One project options definition for
MYVAR
inconan_meson_native.init
orconan_meson_cross.ini
file.
preprocessor_definitions
This attribute allows defining compiler preprocessor definitions, for multiple configurations (Debug, Release, etc).
def generate(self):
tc = MesonToolchain(self)
tc.preprocessor_definitions["MYDEF"] = "MyValue"
tc.generate()
This will be translated to:
One preprocessor definition for
MYDEF
inconan_meson_native.ini
orconan_meson_cross.ini
file.
Generators
The MesonToolchain
only works with the PkgConfigDeps
generator.
Please, do not use other generators, as they can have overlapping definitions that can conflict.
Using the toolchain in developer flow
One of the advantages of using Conan toolchains is that they can help to achieve the exact same build with local development flows, than when the package is created in the cache.
With the MesonToolchain
it is possible to do:
# Lets start in the folder containing the conanfile.py
$ mkdir build && cd build
# Install both debug and release deps and create the toolchain
$ conan install ..
# the build type Release is encoded in the toolchain already.
# This conan_meson_native.iniis specific for release
$ meson setup --native-file conan_meson_native.ini build .
$ meson compile -C build
conf
MesonToolchain
is affected by these [conf] variables:
tools.meson.mesontoolchain:backend
. the meson backend to use. Possible values:ninja
,vs
,vs2010
,vs2015
,vs2017
,vs2019
,xcode
.tools.apple:sdk_path
argument for SDK path in case of Apple cross-compilation. It will be used as value of the flag-isysroot
.tools.android:ndk_path
argument for NDK path in case of Android cross-compilation. It will be used to get some binaries likec
,cpp
andar
used in[binaries]
section from conan_meson_cross.ini.
Apart from that, since Conan 1.47, you can inject extra flags thanks to these ones:
tools.build:cxxflags
list of extra C++ flags that will be used bycpp_args
.tools.build:cflags
list of extra of pure C flags that will be used byc_args
.tools.build:sharedlinkflags
list of extra linker flags that will be used byc_link_args
andcpp_link_args
.tools.build:exelinkflags
list of extra linker flags that will be used byc_link_args
andcpp_link_args
.
Cross-building for Apple and Android
It deserves a special mention because MesonToolchain
is automatically adding all the flags needed
to cross-compile for Apple (MacOS M1, iOS, etc.) and Android.
Apple
It’ll add link flags like -arch XXX
, -isysroot [SDK_PATH]
and the minimum deployment target flag, e.g., -mios-version-min=8.0
into Meson c_args
, c_link_args
, cpp_args
and cpp_link_args
built-in options.
Android
It’ll initialize the c
, cpp
and ar
variables which are needed to cross-compile for Android. For instance:
c == $TOOLCHAIN/bin/llvm-ar
cpp == $TOOLCHAIN/bin/$TARGET$API-clang
ar == $TOOLCHAIN/bin/$TARGET$API-clang++
Where:
$TOOLCHAIN
:[NDK_PATH]/toolchains/llvm/prebuilt/[OS_BUILD]-x86_64/bin
.$TARGET
: target triple, e.g., forarmv8
will beaarch64-linux-android
.$API
: Android API version.
Besides that, you’ll always be able to change any of these variables before being applied thanks
to the MesonToolchain
class interface. For instance:
from conan import ConanFile
from conan.tools.meson import MesonToolchain
class App(ConanFile):
settings = "os", "arch", "compiler", "build_type"
requires = "hello/0.1"
options = {"shared": [True, False]}
default_options = {"shared": False}
def generate(self):
tc = MesonToolchain(self)
tc.cpp = "/path/to/other/compiler"
tc.generate()