meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
# -*- mode: meson -*-
|
|
|
|
|
|
|
|
project('systemd', 'c',
|
|
|
|
version : '233',
|
|
|
|
license : 'LGPLv2+',
|
|
|
|
default_options: [
|
|
|
|
'c_std=gnu99',
|
|
|
|
'prefix=/usr',
|
|
|
|
'sysconfdir=/etc',
|
|
|
|
'localstatedir=/var',
|
|
|
|
],
|
|
|
|
meson_version : '>= 0.39.1',
|
|
|
|
)
|
|
|
|
|
|
|
|
# We need the same data in three different formats, ugh!
|
|
|
|
# Also, for hysterical reasons, we use different variable
|
|
|
|
# names, sometimes. Not all variables are included in every
|
|
|
|
# set. Ugh, ugh, ugh!
|
|
|
|
conf = configuration_data()
|
|
|
|
conf.set_quoted('PACKAGE_STRING', meson.project_name() + ' ' + meson.project_version())
|
|
|
|
conf.set_quoted('PACKAGE_VERSION', meson.project_version())
|
|
|
|
|
|
|
|
substs = configuration_data()
|
|
|
|
substs.set('PACKAGE_URL', 'https://www.freedesktop.org/wiki/Software/systemd')
|
|
|
|
substs.set('PACKAGE_VERSION', meson.project_version())
|
|
|
|
|
|
|
|
m4_defines = []
|
|
|
|
|
|
|
|
#####################################################################
|
|
|
|
|
|
|
|
if get_option('split-usr')
|
|
|
|
conf.set('HAVE_SPLIT_USR', 1)
|
|
|
|
rootprefixdir = '/'
|
|
|
|
else
|
|
|
|
rootprefixdir = '/usr'
|
|
|
|
endif
|
|
|
|
|
|
|
|
sysvinit_path = get_option('sysvinit-path')
|
|
|
|
sysvrcnd_path = get_option('sysvrcnd-path')
|
|
|
|
if sysvinit_path != '' or sysvrcnd_path != ''
|
|
|
|
conf.set('HAVE_SYSV_COMPAT', 1,
|
|
|
|
description : 'SysV init scripts and rcN.d links are supported')
|
|
|
|
endif
|
|
|
|
|
|
|
|
# join_paths ignore the preceding arguments if an absolute component is
|
|
|
|
# encountered, so this should canonicalize various paths when they are
|
|
|
|
# absolute or relative.
|
|
|
|
prefixdir = get_option('prefix')
|
|
|
|
if not prefixdir.startswith('/')
|
|
|
|
error('Prefix is not absolute: "@0@"'.format(prefixdir))
|
|
|
|
endif
|
|
|
|
bindir = join_paths(prefixdir, get_option('bindir'))
|
|
|
|
libdir = join_paths(prefixdir, get_option('libdir'))
|
|
|
|
sysconfdir = join_paths(prefixdir, get_option('sysconfdir'))
|
|
|
|
includedir = join_paths(prefixdir, get_option('includedir'))
|
|
|
|
datadir = join_paths(prefixdir, get_option('datadir'))
|
|
|
|
localstatedir = join_paths('/', get_option('localstatedir'))
|
|
|
|
|
|
|
|
rootbindir = join_paths(rootprefixdir, 'bin')
|
|
|
|
rootlibexecdir = join_paths(rootprefixdir, 'lib/systemd')
|
|
|
|
|
|
|
|
rootlibdir = get_option('rootlibdir')
|
|
|
|
if rootlibdir == ''
|
|
|
|
rootlibdir = join_paths(rootprefixdir, libdir.split('/')[-1])
|
|
|
|
endif
|
|
|
|
|
|
|
|
# Dirs of external packages
|
|
|
|
pkgconfigdatadir = datadir + '/pkgconfig'
|
|
|
|
pkgconfiglibdir = libdir + '/pkgconfig'
|
|
|
|
polkitpolicydir = datadir + '/polkit-1/actions'
|
|
|
|
polkitrulesdir = datadir + '/polkit-1/rules.d'
|
|
|
|
polkitpkladir = localstatedir + '/lib/polkit-1/localauthority/10-vendor.d'
|
|
|
|
varlogdir = localstatedir + '/log'
|
|
|
|
xinitrcdir = sysconfdir + '/X11/xinit/xinitrc.d'
|
|
|
|
rpmmacrosdir = get_option('rpmmacrosdir')
|
|
|
|
|
|
|
|
# Our own paths
|
|
|
|
pkgdatadir = datadir + '/systemd'
|
|
|
|
environmentdir = prefixdir + '/lib/environment.d'
|
|
|
|
pkgsysconfdir = sysconfdir + '/systemd'
|
|
|
|
userunitdir = prefixdir + '/lib/systemd/user'
|
|
|
|
userpresetdir = prefixdir + '/lib/systemd/user-preset'
|
|
|
|
tmpfilesdir = prefixdir + '/lib/tmpfiles.d'
|
|
|
|
sysusersdir = prefixdir + '/lib/sysusers.d'
|
|
|
|
sysctldir = prefixdir + '/lib/sysctl.d'
|
|
|
|
binfmtdir = prefixdir + '/lib/binfmt.d'
|
|
|
|
modulesloaddir = prefixdir + '/lib/modules-load.d'
|
|
|
|
networkdir = rootprefixdir + '/lib/systemd/network'
|
|
|
|
pkgincludedir = includedir + '/systemd'
|
|
|
|
systemgeneratordir = rootlibexecdir + '/system-generators'
|
|
|
|
usergeneratordir = prefixdir + '/lib/systemd/user-generators'
|
|
|
|
systemenvgeneratordir = prefixdir + '/lib/systemd/system-environment-generators'
|
|
|
|
userenvgeneratordir = prefixdir + '/lib/systemd/user-environment-generators'
|
|
|
|
systemshutdowndir = rootlibexecdir + '/system-shutdown'
|
|
|
|
systemsleepdir = rootlibexecdir + '/system-sleep'
|
|
|
|
systemunitdir = rootprefixdir + '/lib/systemd/system'
|
|
|
|
systempresetdir = rootprefixdir + '/lib/systemd/system-preset'
|
|
|
|
udevlibexecdir = rootprefixdir + '/lib/udev'
|
|
|
|
udevhomedir = udevlibexecdir + ''
|
|
|
|
udevrulesdir = udevlibexecdir + '/rules.d'
|
|
|
|
udevhwdbdir = udevlibexecdir + '/hwdb.d'
|
|
|
|
catalogdir = prefixdir + '/lib/systemd/catalog'
|
|
|
|
kernelinstalldir = prefixdir + '/lib/kernel/install.d'
|
|
|
|
factorydir = datadir + '/factory'
|
|
|
|
docdir = datadir + '/doc/systemd'
|
|
|
|
bootlibdir = prefixdir + '/lib/systemd/boot/efi'
|
|
|
|
testsdir = prefixdir + '/lib/systemd/tests'
|
|
|
|
systemdstatedir = localstatedir + '/lib/systemd'
|
|
|
|
catalogstatedir = systemdstatedir + '/catalog'
|
|
|
|
|
|
|
|
dbuspolicydir = get_option('dbuspolicydir')
|
|
|
|
if dbuspolicydir == ''
|
|
|
|
dbuspolicydir = datadir + '/dbus-1/system.d'
|
|
|
|
endif
|
|
|
|
|
|
|
|
dbussessionservicedir = get_option('dbussessionservicedir')
|
|
|
|
if dbussessionservicedir == ''
|
|
|
|
dbussessionservicedir = datadir + '/dbus-1/services'
|
|
|
|
endif
|
|
|
|
|
|
|
|
dbussystemservicedir = get_option('dbussystemservicedir')
|
|
|
|
if dbussystemservicedir == ''
|
|
|
|
dbussystemservicedir = datadir + '/dbus-1/system-services'
|
|
|
|
endif
|
|
|
|
|
|
|
|
pamlibdir = get_option('pamlibdir')
|
|
|
|
if pamlibdir == ''
|
|
|
|
pamlibdir = rootlibdir + '/security'
|
|
|
|
endif
|
|
|
|
|
|
|
|
pamconfdir = get_option('pamconfdir')
|
|
|
|
if pamconfdir == ''
|
|
|
|
pamconfdir = sysconfdir + '/pam.d'
|
|
|
|
endif
|
|
|
|
|
|
|
|
conf.set_quoted('PKGSYSCONFDIR', pkgsysconfdir)
|
|
|
|
conf.set_quoted('SYSTEM_CONFIG_UNIT_PATH', pkgsysconfdir + '/system')
|
|
|
|
conf.set_quoted('SYSTEM_DATA_UNIT_PATH', systemunitdir)
|
|
|
|
conf.set_quoted('SYSTEM_SYSVINIT_PATH', sysvinit_path)
|
|
|
|
conf.set_quoted('SYSTEM_SYSVRCND_PATH', sysvrcnd_path)
|
|
|
|
conf.set_quoted('USER_CONFIG_UNIT_PATH', pkgsysconfdir + '/user')
|
|
|
|
conf.set_quoted('USER_DATA_UNIT_PATH', userunitdir)
|
|
|
|
conf.set_quoted('CERTIFICATE_ROOT', get_option('certificate-root'))
|
|
|
|
conf.set_quoted('CATALOG_DATABASE', catalogstatedir + '/database')
|
|
|
|
conf.set_quoted('SYSTEMD_CGROUP_AGENT_PATH', rootlibexecdir + '/systemd-cgroups-agent')
|
|
|
|
conf.set_quoted('SYSTEMD_BINARY_PATH', rootlibexecdir + '/systemd')
|
|
|
|
conf.set_quoted('SYSTEMD_FSCK_PATH', rootlibexecdir + '/systemd-fsck')
|
|
|
|
conf.set_quoted('SYSTEMD_SHUTDOWN_BINARY_PATH', rootlibexecdir + '/systemd-shutdown')
|
|
|
|
conf.set_quoted('SYSTEMD_SLEEP_BINARY_PATH', rootlibexecdir + '/systemd-sleep')
|
|
|
|
conf.set_quoted('SYSTEMCTL_BINARY_PATH', rootbindir + '/systemctl')
|
|
|
|
conf.set_quoted('SYSTEMD_TTY_ASK_PASSWORD_AGENT_BINARY_PATH', rootbindir + '/systemd-tty-ask-password-agent')
|
|
|
|
conf.set_quoted('SYSTEMD_STDIO_BRIDGE_BINARY_PATH', bindir + '/systemd-stdio-bridge')
|
|
|
|
conf.set_quoted('ROOTPREFIX', rootprefixdir)
|
|
|
|
conf.set_quoted('RANDOM_SEED_DIR', localstatedir + '/lib/systemd/')
|
|
|
|
conf.set_quoted('RANDOM_SEED', localstatedir + '/lib/systemd/random-seed')
|
|
|
|
conf.set_quoted('SYSTEMD_CRYPTSETUP_PATH', rootlibexecdir + '/systemd-cryptsetup')
|
|
|
|
conf.set_quoted('SYSTEM_GENERATOR_PATH', systemgeneratordir)
|
|
|
|
conf.set_quoted('USER_GENERATOR_PATH', usergeneratordir)
|
|
|
|
conf.set_quoted('SYSTEM_ENV_GENERATOR_PATH', systemenvgeneratordir)
|
|
|
|
conf.set_quoted('USER_ENV_GENERATOR_PATH', userenvgeneratordir)
|
|
|
|
conf.set_quoted('SYSTEM_SHUTDOWN_PATH', systemshutdowndir)
|
|
|
|
conf.set_quoted('SYSTEM_SLEEP_PATH', systemsleepdir)
|
|
|
|
conf.set_quoted('SYSTEMD_KBD_MODEL_MAP', pkgdatadir + '/kbd-model-map')
|
|
|
|
conf.set_quoted('SYSTEMD_LANGUAGE_FALLBACK_MAP', pkgdatadir + '/language-fallback-map')
|
|
|
|
conf.set_quoted('UDEVLIBEXECDIR', udevlibexecdir)
|
|
|
|
conf.set_quoted('POLKIT_AGENT_BINARY_PATH', bindir + '/pkttyagent')
|
|
|
|
conf.set_quoted('LIBDIR', libdir)
|
|
|
|
conf.set_quoted('ROOTLIBDIR', rootlibdir)
|
|
|
|
conf.set_quoted('ROOTLIBEXECDIR', rootlibexecdir)
|
|
|
|
conf.set_quoted('BOOTLIBDIR', bootlibdir)
|
|
|
|
conf.set_quoted('SYSTEMD_PULL_PATH', rootlibexecdir + '/systemd-pull')
|
|
|
|
conf.set_quoted('SYSTEMD_IMPORT_PATH', rootlibexecdir + '/systemd-import')
|
|
|
|
conf.set_quoted('SYSTEMD_EXPORT_PATH', rootlibexecdir + '/systemd-export')
|
|
|
|
conf.set_quoted('VENDOR_KEYRING_PATH', rootlibexecdir + '/import-pubring.gpg')
|
|
|
|
conf.set_quoted('USER_KEYRING_PATH', pkgsysconfdir + '/import-pubring.gpg')
|
|
|
|
conf.set_quoted('DOCUMENT_ROOT', pkgdatadir + '/gatewayd')
|
|
|
|
|
|
|
|
conf.set_quoted('ABS_BUILD_DIR', meson.build_root())
|
|
|
|
conf.set_quoted('ABS_SRC_DIR', meson.source_root())
|
|
|
|
|
|
|
|
substs.set('prefix', prefixdir)
|
|
|
|
substs.set('pkgsysconfdir', pkgsysconfdir)
|
|
|
|
substs.set('rootlibexecdir', rootlibexecdir)
|
|
|
|
substs.set('systemunitdir', systemunitdir)
|
|
|
|
substs.set('userunitdir', userunitdir)
|
|
|
|
substs.set('systempresetdir', systempresetdir)
|
|
|
|
substs.set('userpresetdir', userpresetdir)
|
|
|
|
substs.set('udevhwdbdir', udevhwdbdir)
|
|
|
|
substs.set('udevrulesdir', udevrulesdir)
|
|
|
|
substs.set('catalogdir', catalogdir)
|
|
|
|
substs.set('tmpfilesdir', tmpfilesdir)
|
|
|
|
substs.set('sysusersdir', sysusersdir)
|
|
|
|
substs.set('sysctldir', sysctldir)
|
|
|
|
substs.set('binfmtdir', binfmtdir)
|
|
|
|
substs.set('modulesloaddir', modulesloaddir)
|
|
|
|
substs.set('systemgeneratordir', systemgeneratordir)
|
|
|
|
substs.set('usergeneratordir', usergeneratordir)
|
|
|
|
substs.set('systemenvgeneratordir', systemenvgeneratordir)
|
|
|
|
substs.set('userenvgeneratordir', userenvgeneratordir)
|
|
|
|
substs.set('systemshutdowndir', systemshutdowndir)
|
|
|
|
substs.set('systemsleepdir', systemsleepdir)
|
2017-04-11 00:37:52 +08:00
|
|
|
substs.set('SYSTEMCTL', rootbindir + '/systemctl')
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
|
|
|
|
#####################################################################
|
|
|
|
|
|
|
|
cc = meson.get_compiler('c')
|
|
|
|
pkgconfig = import('pkgconfig')
|
|
|
|
|
|
|
|
foreach arg : ['-Wundef',
|
|
|
|
'-Wlogical-op',
|
|
|
|
'-Wmissing-include-dirs',
|
|
|
|
'-Wold-style-definition',
|
|
|
|
'-Wpointer-arith',
|
|
|
|
'-Winit-self',
|
|
|
|
'-Wdeclaration-after-statement',
|
|
|
|
'-Wfloat-equal',
|
|
|
|
'-Wsuggest-attribute=noreturn',
|
|
|
|
'-Werror=missing-prototypes',
|
|
|
|
'-Werror=implicit-function-declaration',
|
|
|
|
'-Werror=missing-declarations',
|
|
|
|
'-Werror=return-type',
|
|
|
|
'-Werror=incompatible-pointer-types',
|
|
|
|
'-Werror=format=2',
|
|
|
|
'-Wstrict-prototypes',
|
|
|
|
'-Wredundant-decls',
|
|
|
|
'-Wmissing-noreturn',
|
|
|
|
'-Wshadow',
|
|
|
|
'-Wendif-labels',
|
|
|
|
'-Wstrict-aliasing=2',
|
|
|
|
'-Wwrite-strings',
|
|
|
|
'-Wno-unused-parameter',
|
|
|
|
'-Wno-missing-field-initializers',
|
|
|
|
'-Wno-unused-result',
|
|
|
|
'-Wno-format-signedness',
|
|
|
|
'-Werror=overflow',
|
|
|
|
'-Wdate-time',
|
|
|
|
'-Wnested-externs',
|
|
|
|
'-ffast-math',
|
|
|
|
'-fno-common',
|
|
|
|
'-fdiagnostics-show-option',
|
|
|
|
'-fno-strict-aliasing',
|
|
|
|
'-fvisibility=hidden',
|
|
|
|
'-fstack-protector',
|
|
|
|
'-fstack-protector-strong',
|
|
|
|
'-fPIE',
|
|
|
|
'--param=ssp-buffer-size=4',
|
|
|
|
]
|
|
|
|
if cc.has_argument(arg)
|
|
|
|
add_project_arguments(arg, language : 'c')
|
|
|
|
endif
|
|
|
|
endforeach
|
|
|
|
|
|
|
|
if cc.compiles('
|
|
|
|
#include <time.h>
|
|
|
|
#include <inttypes.h>
|
|
|
|
typedef uint64_t usec_t;
|
|
|
|
usec_t now(clockid_t clock);
|
|
|
|
int main(void) {
|
|
|
|
struct timespec now;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
')
|
|
|
|
# TODO: message?
|
|
|
|
add_project_arguments('-Werror=shadow', language : 'c')
|
|
|
|
endif
|
|
|
|
|
|
|
|
if cc.get_id() == 'clang'
|
|
|
|
foreach arg : ['-Wno-typedef-redefinition',
|
|
|
|
'-Wno-gnu-variable-sized-type-not-at-end',
|
|
|
|
]
|
|
|
|
if cc.has_argument(arg)
|
|
|
|
add_project_arguments(arg, language : 'c')
|
|
|
|
endif
|
|
|
|
endforeach
|
|
|
|
endif
|
|
|
|
|
|
|
|
# --as-needed and --no-undefined are provided by meson by default,
|
|
|
|
# run mesonconf to see what is enabled
|
|
|
|
foreach arg : ['-Wl,-z,relro',
|
|
|
|
'-Wl,-z,now',
|
|
|
|
'-pie',
|
|
|
|
'-Wl,-fuse-ld=gold',
|
|
|
|
]
|
|
|
|
if cc.has_argument(arg)
|
|
|
|
add_project_link_arguments(arg, language : 'c')
|
|
|
|
endif
|
|
|
|
endforeach
|
|
|
|
|
|
|
|
#####################################################################
|
|
|
|
# compilation result tests
|
|
|
|
|
|
|
|
conf.set('_GNU_SOURCE', 1)
|
|
|
|
conf.set('__SANE_USERSPACE_TYPES__', 1)
|
|
|
|
|
|
|
|
conf.set('SIZEOF_PID_T', cc.sizeof('pid_t', prefix : '#include <sys/types.h>'))
|
|
|
|
conf.set('SIZEOF_UID_T', cc.sizeof('uid_t', prefix : '#include <sys/types.h>'))
|
|
|
|
conf.set('SIZEOF_GID_T', cc.sizeof('gid_t', prefix : '#include <sys/types.h>'))
|
|
|
|
conf.set('SIZEOF_DEV_T', cc.sizeof('dev_t', prefix : '#include <sys/types.h>'))
|
|
|
|
conf.set('SIZEOF_INO_T', cc.sizeof('ino_t', prefix : '#include <sys/types.h>'))
|
|
|
|
conf.set('SIZEOF_TIME_T', cc.sizeof('time_t', prefix : '#include <sys/time.h>'))
|
|
|
|
conf.set('SIZEOF_RLIM_T', cc.sizeof('rlim_t', prefix : '#include <sys/resource.h>'))
|
|
|
|
|
|
|
|
decl_headers = '''
|
|
|
|
#include <uchar.h>
|
|
|
|
#include <linux/ethtool.h>
|
|
|
|
'''
|
|
|
|
# FIXME: key_serial_t is only defined in keyutils.h, this is bound to fail
|
|
|
|
|
|
|
|
foreach decl : ['char16_t',
|
|
|
|
'char32_t',
|
|
|
|
'key_serial_t',
|
|
|
|
'struct ethtool_link_settings',
|
|
|
|
]
|
|
|
|
if cc.sizeof(decl, prefix : decl_headers) > 0
|
|
|
|
# We get -1 if the size cannot be determined
|
|
|
|
conf.set('HAVE_' + decl.underscorify().to_upper(), 1)
|
|
|
|
endif
|
|
|
|
endforeach
|
|
|
|
|
|
|
|
foreach decl : [['IFLA_INET6_ADDR_GEN_MODE', 'linux/if_link.h'],
|
|
|
|
['IN6_ADDR_GEN_MODE_STABLE_PRIVACY', 'linux/if_link.h'],
|
|
|
|
['IFLA_VRF_TABLE', 'linux/if_link.h'],
|
|
|
|
['IFLA_MACVLAN_FLAGS', 'linux/if_link.h'],
|
|
|
|
['IFLA_IPVLAN_MODE', 'linux/if_link.h'],
|
|
|
|
['IFLA_PHYS_PORT_ID', 'linux/if_link.h'],
|
|
|
|
['IFLA_BOND_AD_INFO', 'linux/if_link.h'],
|
|
|
|
['IFLA_VLAN_PROTOCOL', 'linux/if_link.h'],
|
|
|
|
['IFLA_VXLAN_REMCSUM_NOPARTIAL', 'linux/if_link.h'],
|
|
|
|
['IFLA_VXLAN_GPE', 'linux/if_link.h'],
|
|
|
|
# if_tunnel.h is buggy and cannot be included on its own
|
|
|
|
['IFLA_VTI_REMOTE', 'linux/if_tunnel.h', '#include <net/if.h>'],
|
|
|
|
['IFLA_IPTUN_ENCAP_DPORT', 'linux/if_tunnel.h', '#include <net/if.h>'],
|
|
|
|
['IFLA_GRE_ENCAP_DPORT', 'linux/if_tunnel.h', '#include <net/if.h>'],
|
|
|
|
['IFLA_BRIDGE_VLAN_INFO', 'linux/if_bridge.h'],
|
|
|
|
['IFLA_BRPORT_PROXYARP', 'linux/if_link.h'],
|
|
|
|
['IFLA_BRPORT_LEARNING_SYNC', 'linux/if_link.h'],
|
|
|
|
['IFLA_BR_VLAN_DEFAULT_PVID', 'linux/if_link.h'],
|
|
|
|
['NDA_IFINDEX', 'linux/neighbour.h'],
|
|
|
|
['IFA_FLAGS', 'linux/if_addr.h'],
|
|
|
|
['LO_FLAGS_PARTSCAN', 'linux/loop.h'],
|
|
|
|
]
|
|
|
|
prefix = decl.length() > 2 ? decl[2] : ''
|
|
|
|
have = cc.has_header_symbol(decl[1], decl[0], prefix : prefix)
|
|
|
|
conf.set10('HAVE_DECL_' + decl[0], have)
|
|
|
|
endforeach
|
|
|
|
|
|
|
|
skip = false
|
|
|
|
foreach ident : ['secure_getenv', '__secure_getenv']
|
|
|
|
if not skip and cc.has_function(ident)
|
|
|
|
conf.set('HAVE_' + ident.to_upper(), 1)
|
|
|
|
skip = true
|
|
|
|
endif
|
|
|
|
endforeach
|
|
|
|
|
|
|
|
foreach ident : [
|
|
|
|
['memfd_create', '''#include <sys/memfd.h>'''],
|
|
|
|
['gettid', '''#include <sys/types.h>'''],
|
|
|
|
['pivot_root', '''#include <stdlib.h>'''], # no known header declares pivot_root
|
|
|
|
['name_to_handle_at', '''#define _GNU_SOURCE
|
|
|
|
#include <sys/types.h>
|
|
|
|
#include <sys/stat.h>
|
|
|
|
#include <fcntl.h>'''],
|
|
|
|
['setns', '''#define _GNU_SOURCE
|
|
|
|
#include <sched.h>'''],
|
|
|
|
['getrandom', '''#include <sys/random.h>'''],
|
|
|
|
['renameat2', '''#include <stdio.h>'''],
|
|
|
|
['kcmp', '''#include <linux/kcmp.h>'''],
|
|
|
|
['keyctl', '''#include <sys/types.h>
|
|
|
|
#include <keyutils.h>'''],
|
|
|
|
['copy_file_range', '''#include <sys/syscall.h>
|
|
|
|
#include <unistd.h>'''],
|
|
|
|
['explicit_bzero' , '''#include <strings.h>'''],
|
|
|
|
]
|
|
|
|
|
|
|
|
have = cc.has_function(ident[0], prefix : ident[1])
|
|
|
|
conf.set10('HAVE_DECL_' + ident[0].to_upper(), have)
|
|
|
|
endforeach
|
|
|
|
|
|
|
|
#####################################################################
|
|
|
|
|
|
|
|
sed = find_program('sed')
|
|
|
|
grep = find_program('grep')
|
|
|
|
awk = find_program('awk')
|
2017-04-05 12:40:21 +08:00
|
|
|
m4 = find_program('/usr/bin/m4')
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
stat = find_program('stat')
|
2017-04-07 02:33:27 +08:00
|
|
|
git = find_program('git', required : false)
|
|
|
|
etags = find_program('etags', required : false)
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
|
2017-04-10 11:55:50 +08:00
|
|
|
meson_make_symlink = meson.source_root() + '/tools/meson-make-symlink.sh'
|
2017-04-10 11:55:05 +08:00
|
|
|
mkdir_p = 'mkdir -p $DESTDIR/@0@'
|
|
|
|
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
# if -Dxxx-path option is found, use that. Otherwise, check in $PATH,
|
|
|
|
# /usr/sbin, /sbin, and fall back to the default from middle column.
|
|
|
|
progs = [['telinit', '/lib/sysvinit/telinit'],
|
|
|
|
['quotaon', '/usr/sbin/quotaon' ],
|
|
|
|
['quotacheck', '/usr/sbin/quotacheck' ],
|
|
|
|
['kill', '/usr/bin/kill' ],
|
|
|
|
['kmod', '/usr/bin/kmod' ],
|
|
|
|
['kexec', '/usr/sbin/kexec' ],
|
|
|
|
['sulogin', '/usr/sbin/sulogin' ],
|
|
|
|
['mount', '/usr/bin/mount', 'MOUNT_PATH'],
|
|
|
|
['umount', '/usr/bin/umount', 'UMOUNT_PATH'],
|
|
|
|
['loadkeys', '/usr/bin/loadkeys', 'KBD_LOADKEYS'],
|
|
|
|
['setfont', '/usr/bin/setfont', 'KBD_SETFONT'],
|
|
|
|
]
|
|
|
|
foreach prog : progs
|
|
|
|
path = get_option(prog[0] + '-path')
|
|
|
|
if path != ''
|
|
|
|
message('Using @1@ for @0@'.format(prog[0], path))
|
|
|
|
else
|
|
|
|
exe = find_program(prog[0], '/usr/sbin/' + prog[0], '/sbin/' + prog[0], required: false)
|
|
|
|
path = exe.found() ? exe.path() : prog[1]
|
|
|
|
endif
|
|
|
|
name = prog.length() > 2 ? prog[2] : prog[0].to_upper()
|
|
|
|
conf.set_quoted(name, path)
|
2017-04-11 00:37:52 +08:00
|
|
|
substs.set(name, path)
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
endforeach
|
|
|
|
|
|
|
|
# TODO: add ln --relative check
|
|
|
|
# AS_IF([! ln --relative --help > /dev/null 2>&1], [AC_MSG_ERROR([*** ln doesn't support --relative ***])])
|
|
|
|
|
|
|
|
############################################################
|
|
|
|
|
|
|
|
gperf = find_program('gperf')
|
|
|
|
|
|
|
|
gperf_test_format = '''
|
|
|
|
#include <string.h>
|
|
|
|
const char * in_word_set(const char *, @0@);
|
|
|
|
@1@
|
|
|
|
'''
|
|
|
|
gperf_snippet_format = 'echo foo,bar | @0@ -L ANSI-C'
|
|
|
|
gperf_snippet = run_command('sh', '-c', gperf_snippet_format.format(gperf.path()))
|
|
|
|
gperf_test = gperf_test_format.format('size_t', gperf_snippet.stdout())
|
|
|
|
if cc.compiles(gperf_test)
|
|
|
|
gperf_len_type = 'size_t'
|
|
|
|
else
|
|
|
|
gperf_test = gperf_test_format.format('unsigned', gperf_snippet.stdout())
|
|
|
|
if cc.compiles(gperf_test)
|
|
|
|
gperf_len_type = 'unsigned'
|
|
|
|
else
|
|
|
|
error('unable to determine gperf len type')
|
|
|
|
endif
|
|
|
|
endif
|
|
|
|
message('gperf len type is @0@'.format(gperf_len_type))
|
|
|
|
conf.set('GPERF_LEN_TYPE', gperf_len_type, description : 'The type of gperf "len" parameter')
|
|
|
|
|
|
|
|
############################################################
|
|
|
|
|
|
|
|
if not cc.has_header('sys/capability.h')
|
|
|
|
error('POSIX caps headers not found')
|
|
|
|
endif
|
|
|
|
foreach header : ['linux/btrfs.h',
|
|
|
|
'linux/memfd.h',
|
|
|
|
'linux/vm_sockets.h',
|
|
|
|
'valgrind/memcheck.h',
|
|
|
|
'valgrind/valgrind.h',
|
|
|
|
]
|
|
|
|
if cc.has_header(header)
|
|
|
|
conf.set('HAVE_' + header.underscorify().to_upper(), 1)
|
|
|
|
endif
|
|
|
|
endforeach
|
|
|
|
|
|
|
|
############################################################
|
|
|
|
|
|
|
|
conf.set_quoted('FALLBACK_HOSTNAME', get_option('fallback-hostname'))
|
|
|
|
|
|
|
|
default_hierarchy = get_option('default-hierarchy')
|
|
|
|
conf.set_quoted('DEFAULT_HIERARCHY_NAME', default_hierarchy,
|
|
|
|
description : 'default cgroup hierarchy as string')
|
|
|
|
if default_hierarchy == 'legacy'
|
|
|
|
conf.set('DEFAULT_HIERARCHY', 'CGROUP_UNIFIED_NONE')
|
|
|
|
elif default_hierarchy == 'hybrid'
|
|
|
|
conf.set('DEFAULT_HIERARCHY', 'CGROUP_UNIFIED_SYSTEMD')
|
|
|
|
else
|
|
|
|
conf.set('DEFAULT_HIERARCHY', 'CGROUP_UNIFIED_ALL')
|
|
|
|
endif
|
|
|
|
|
|
|
|
time_epoch = get_option('time-epoch')
|
|
|
|
if time_epoch == ''
|
|
|
|
NEWS = files('NEWS')
|
|
|
|
time_epoch = run_command(stat, '-c', '%Y', NEWS).stdout()
|
|
|
|
endif
|
|
|
|
time_epoch = time_epoch.to_int()
|
|
|
|
conf.set('TIME_EPOCH', time_epoch)
|
|
|
|
|
|
|
|
system_uid_max = get_option('system-uid-max')
|
|
|
|
if system_uid_max == ''
|
|
|
|
system_uid_max = run_command(
|
|
|
|
awk,
|
|
|
|
'BEGIN { uid=999 } /^\s*SYS_UID_MAX\s+/ { uid=$2 } END { print uid }',
|
|
|
|
'/etc/login.defs').stdout()
|
|
|
|
endif
|
|
|
|
system_uid_max = system_uid_max.to_int()
|
|
|
|
conf.set('SYSTEM_UID_MAX', system_uid_max)
|
|
|
|
substs.set('systemuidmax', system_uid_max)
|
|
|
|
message('Maximum system UID is @0@'.format(system_uid_max))
|
|
|
|
|
|
|
|
conf.set_quoted('NOBODY_USER_NAME', get_option('nobody-user'))
|
|
|
|
conf.set_quoted('NOBODY_GROUP_NAME', get_option('nobody-group'))
|
|
|
|
|
|
|
|
system_gid_max = get_option('system-gid-max')
|
|
|
|
if system_gid_max == ''
|
|
|
|
system_gid_max = run_command(
|
|
|
|
awk,
|
|
|
|
'BEGIN { gid=999 } /^\s*SYS_GID_MAX\s+/ { gid=$2 } END { print gid }',
|
|
|
|
'/etc/login.defs').stdout()
|
|
|
|
endif
|
|
|
|
system_gid_max = system_gid_max.to_int()
|
|
|
|
conf.set('SYSTEM_GID_MAX', system_gid_max)
|
|
|
|
substs.set('systemgidmax', system_gid_max)
|
|
|
|
message('Maximum system GID is @0@'.format(system_gid_max))
|
|
|
|
|
|
|
|
tty_gid = get_option('tty-gid')
|
|
|
|
conf.set('TTY_GID', tty_gid)
|
|
|
|
m4_defines += ['-DTTY_GID=' + tty_gid]
|
|
|
|
|
|
|
|
if get_option('adm-group')
|
|
|
|
m4_defines += ['-DENABLE_ADM_GROUP']
|
|
|
|
endif
|
|
|
|
|
|
|
|
if get_option('wheel-group')
|
|
|
|
m4_defines += ['-DENABLE_WHEEL_GROUP']
|
|
|
|
endif
|
|
|
|
|
|
|
|
substs.set('DEV_KVM_MODE', get_option('dev-kvm-mode'))
|
|
|
|
|
|
|
|
conf.set10('KILL_USER_PROCESSES', get_option('default-kill-user-processes'))
|
|
|
|
|
|
|
|
default_dnssec = get_option('default-dnssec')
|
|
|
|
conf.set('DEFAULT_DNSSEC_MODE',
|
|
|
|
'DNSSEC_' + default_dnssec.underscorify().to_upper())
|
|
|
|
|
|
|
|
conf.set_quoted('DNS_SERVERS', get_option('dns-servers'))
|
|
|
|
|
|
|
|
conf.set_quoted('NTP_SERVERS', get_option('ntp-servers'))
|
|
|
|
|
|
|
|
conf.set_quoted('GETTEXT_PACKAGE', meson.project_name())
|
|
|
|
|
|
|
|
#####################################################################
|
|
|
|
|
|
|
|
threads = dependency('threads')
|
|
|
|
librt = cc.find_library('rt')
|
|
|
|
libm = cc.find_library('m')
|
|
|
|
libdl = cc.find_library('dl')
|
|
|
|
libcrypt = cc.find_library('crypt')
|
|
|
|
|
|
|
|
libcap = dependency('libcap')
|
|
|
|
libmount = dependency('mount',
|
|
|
|
version : '>= 2.27')
|
|
|
|
|
|
|
|
want_seccomp = get_option('seccomp')
|
|
|
|
if want_seccomp != 'no'
|
|
|
|
libseccomp = dependency('libseccomp',
|
|
|
|
required : want_seccomp == 'yes')
|
|
|
|
if libseccomp.found()
|
|
|
|
conf.set('HAVE_SECCOMP', 1)
|
|
|
|
m4_defines += ['-DHAVE_SECCOMP']
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libseccomp = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_selinux = get_option('selinux')
|
|
|
|
if want_selinux != 'no'
|
|
|
|
libselinux = dependency('libselinux',
|
|
|
|
version : '>= 2.1.9',
|
|
|
|
required : want_selinux == 'yes')
|
|
|
|
if libselinux.found()
|
|
|
|
conf.set('HAVE_SELINUX', 1)
|
|
|
|
m4_defines += ['-DHAVE_SELINUX']
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libselinux = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_apparmor = get_option('apparmor')
|
|
|
|
if want_apparmor != 'no'
|
|
|
|
libapparmor = dependency('libapparmor',
|
|
|
|
required : want_apparmor == 'yes')
|
|
|
|
if libapparmor.found()
|
|
|
|
conf.set('HAVE_APPARMOR', 1)
|
|
|
|
m4_defines += ['-DHAVE_APPARMOR']
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libapparmor = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_smack = get_option('smack')
|
|
|
|
if want_smack != 'no'
|
|
|
|
libsmack = dependency('libsmack',
|
|
|
|
required : want_smack == 'yes')
|
|
|
|
if libsmack.found()
|
|
|
|
conf.set('HAVE_SMACK', 1)
|
|
|
|
m4_defines += ['-DHAVE_SMACK']
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libsmack = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
smack_run_label = get_option('smack-run-label')
|
|
|
|
if smack_run_label != ''
|
|
|
|
conf.set_quoted('SMACK_RUN_LABEL', smack_run_label)
|
|
|
|
m4_defines += ['-DHAVE_SMACK_RUN_LABEL']
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_audit = get_option('audit')
|
|
|
|
if want_audit != 'no'
|
|
|
|
libaudit = dependency('audit', required : want_audit == 'yes')
|
|
|
|
if libaudit.found()
|
|
|
|
conf.set('HAVE_AUDIT', 1)
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libaudit = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_blkid = get_option('blkid')
|
|
|
|
if want_blkid != 'no'
|
|
|
|
libblkid = dependency('blkid', required : want_blkid == 'yes')
|
|
|
|
if libblkid.found()
|
|
|
|
conf.set('HAVE_BLKID', 1)
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libblkid = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_kmod = get_option('kmod')
|
|
|
|
if want_kmod != 'no'
|
|
|
|
libkmod = dependency('libkmod',
|
|
|
|
version : '>= 15',
|
|
|
|
required : want_kmod == 'yes')
|
|
|
|
if libkmod.found()
|
|
|
|
conf.set('HAVE_KMOD', 1)
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libkmod = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_pam = get_option('pam')
|
|
|
|
if want_pam != 'no'
|
|
|
|
libpam = cc.find_library('pam', required : want_pam == 'yes')
|
|
|
|
libpam_misc = cc.find_library('pam_misc', required : want_pam == 'yes')
|
|
|
|
if libpam.found() and libpam_misc.found()
|
|
|
|
conf.set('HAVE_PAM', 1)
|
|
|
|
m4_defines += ['-DHAVE_PAM']
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libpam = []
|
|
|
|
libpam_misc = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_microhttpd = get_option('microhttpd')
|
|
|
|
if want_microhttpd != 'no'
|
|
|
|
libmicrohttpd = dependency('libmicrohttpd',
|
|
|
|
version : '>= 0.9.33',
|
|
|
|
required : want_microhttpd == 'yes')
|
|
|
|
if libmicrohttpd.found()
|
|
|
|
conf.set('HAVE_MICROHTTPD', 1)
|
|
|
|
m4_defines += ['-DHAVE_MICROHTTPD']
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libmicrohttpd = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_libcryptsetup = get_option('libcryptsetup')
|
|
|
|
if want_libcryptsetup != 'no'
|
|
|
|
libcryptsetup = dependency('libcryptsetup',
|
|
|
|
version : '>= 1.6.0',
|
|
|
|
required : want_libcryptsetup == 'yes')
|
|
|
|
if libcryptsetup.found()
|
|
|
|
conf.set('HAVE_LIBCRYPTSETUP', 1)
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libcryptsetup = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_libcurl = get_option('libcurl')
|
|
|
|
if want_libcurl != 'no'
|
|
|
|
libcurl = dependency('libcurl',
|
|
|
|
version : '>= 7.32.0',
|
|
|
|
required : want_libcurl == 'yes')
|
|
|
|
if libcurl.found()
|
|
|
|
conf.set('HAVE_LIBCURL', 1)
|
|
|
|
m4_defines += ['-DHAVE_LIBCURL']
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libcurl = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_libidn = get_option('libidn')
|
|
|
|
if want_libidn != 'no'
|
|
|
|
libidn = dependency('libidn',
|
|
|
|
required : want_libidn == 'yes')
|
|
|
|
if libidn.found()
|
|
|
|
conf.set('HAVE_LIBIDN', 1)
|
|
|
|
m4_defines += ['-DHAVE_LIBIDN']
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libidn = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_libiptc = get_option('libiptc')
|
|
|
|
if want_libiptc != 'no'
|
|
|
|
libiptc = dependency('libiptc',
|
|
|
|
required : want_libiptc == 'yes')
|
|
|
|
if libiptc.found()
|
|
|
|
conf.set('HAVE_LIBIPTC', 1)
|
|
|
|
m4_defines += ['-DHAVE_LIBIPTC']
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libiptc = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_qrencode = get_option('qrencode')
|
|
|
|
if want_qrencode != 'no'
|
|
|
|
libqrencode = dependency('libqrencode',
|
|
|
|
required : want_qrencode == 'yes')
|
|
|
|
if libqrencode.found()
|
|
|
|
conf.set('HAVE_QRENCODE', 1)
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libqrencode = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_gnutls = get_option('gnutls')
|
|
|
|
if want_gnutls != 'no'
|
|
|
|
libgnutls = dependency('gnutls',
|
|
|
|
version : '>= 3.1.4',
|
|
|
|
required : want_gnutls == 'yes')
|
|
|
|
if libgnutls.found()
|
|
|
|
conf.set('HAVE_GNUTLS', 1)
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libgnutls = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_elfutils = get_option('elfutils')
|
|
|
|
if want_elfutils != 'no'
|
|
|
|
libdw = dependency('libdw',
|
|
|
|
required : want_elfutils == 'yes')
|
|
|
|
if libdw.found()
|
|
|
|
conf.set('HAVE_ELFUTILS', 1)
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libdw = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_zlib = get_option('zlib')
|
|
|
|
if want_zlib != 'no'
|
|
|
|
libz = dependency('zlib',
|
|
|
|
required : want_zlib == 'yes')
|
|
|
|
if libz.found()
|
|
|
|
conf.set('HAVE_ZLIB', 1)
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libz = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_bzip2 = get_option('bzip2')
|
|
|
|
if want_bzip2 != 'no'
|
|
|
|
libbzip2 = cc.find_library('bz2',
|
|
|
|
required : want_bzip2 == 'yes')
|
|
|
|
if libbzip2.found()
|
|
|
|
conf.set('HAVE_BZIP2', 1)
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libbzip2 = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_xz = get_option('xz')
|
|
|
|
if want_xz != 'no'
|
|
|
|
libxz = dependency('liblzma',
|
|
|
|
required : want_xz == 'yes')
|
|
|
|
if libxz.found()
|
|
|
|
conf.set('HAVE_XZ', 1)
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libxz = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_lz4 = get_option('lz4')
|
|
|
|
if want_lz4 != 'no'
|
|
|
|
liblz4 = dependency('liblz4',
|
|
|
|
required : want_lz4 == 'yes')
|
|
|
|
if liblz4.found()
|
|
|
|
conf.set('HAVE_LZ4', 1)
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
liblz4 = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
libacl = cc.find_library('acl', required : false)
|
|
|
|
if libacl.found()
|
|
|
|
conf.set('HAVE_ACL', 1)
|
|
|
|
m4_defines += ['-DHAVE_ACL']
|
|
|
|
endif
|
|
|
|
|
2017-04-07 12:19:09 +08:00
|
|
|
want_glib = get_option('glib')
|
|
|
|
if want_glib != 'no'
|
|
|
|
libglib = dependency('glib-2.0',
|
|
|
|
version : '>= 2.22.0',
|
|
|
|
required : want_glib == 'yes')
|
|
|
|
libgobject = dependency('gobject-2.0',
|
|
|
|
version : '>= 2.22.0',
|
|
|
|
required : want_glib == 'yes')
|
|
|
|
libgio = dependency('gio-2.0',
|
|
|
|
required : want_glib == 'yes')
|
|
|
|
if libglib.found() and libgobject.found() and libgio.found()
|
|
|
|
conf.set('HAVE_GLIB', 1)
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libglib = []
|
|
|
|
libgobject = []
|
|
|
|
libgio = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_dbus = get_option('dbus')
|
|
|
|
if want_dbus != 'no'
|
|
|
|
libdbus = dependency('dbus-1',
|
|
|
|
version : '>= 1.3.2',
|
|
|
|
required : want_dbus == 'yes')
|
|
|
|
if libdbus.found()
|
|
|
|
conf.set('HAVE_DBUS', 1)
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libdbus = []
|
|
|
|
endif
|
|
|
|
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
want_libgcrypt = get_option('libgcrypt')
|
|
|
|
if want_libgcrypt != 'no'
|
|
|
|
libgcrypt = cc.find_library('gcrypt', required : want_libgcrypt == 'yes')
|
|
|
|
if libgcrypt.found()
|
|
|
|
conf.set('HAVE_LIBGCRYPT', 1)
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
libgcrypt = []
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_importd = get_option('importd')
|
|
|
|
if want_importd != 'no'
|
|
|
|
have_deps = (conf.get('HAVE_LIBCURL', 0) == 1 and
|
|
|
|
conf.get('HAVE_ZLIB', 0) == 1 and
|
|
|
|
conf.get('HAVE_BZIP2', 0) == 1 and
|
|
|
|
conf.get('HAVE_XZ', 0) == 1 and
|
|
|
|
conf.get('HAVE_LZ4', 0) == 1)
|
|
|
|
if have_deps
|
|
|
|
conf.set('ENABLE_IMPORTD', 1)
|
|
|
|
elif want_importd == 'yes'
|
|
|
|
error('importd support was requested, but dependencies are not available')
|
|
|
|
endif
|
|
|
|
endif
|
|
|
|
|
|
|
|
want_remote = get_option('remote')
|
|
|
|
if want_remote != 'no'
|
|
|
|
have_deps = [conf.get('HAVE_MICROHTTPD', 0) == 1,
|
|
|
|
conf.get('HAVE_LIBCURL', 0) == 1]
|
|
|
|
# sd-j-remote requires µhttpd, and sd-j-upload requires libcurl, so
|
|
|
|
# it's possible to build one without the other. Complain only if
|
|
|
|
# support was explictly requested. The auxiliary files like sysusers
|
|
|
|
# config should be installed when any of the programs are built.
|
|
|
|
if want_remote == 'yes' and not (have_deps[0] and have_deps[1])
|
|
|
|
error('remote support was requested, but dependencies are not available')
|
|
|
|
endif
|
|
|
|
if have_deps[0] or have_deps[1]
|
|
|
|
conf.set('ENABLE_REMOTE', 1)
|
|
|
|
endif
|
|
|
|
endif
|
|
|
|
|
|
|
|
foreach pair : [['utmp', 'HAVE_UTMP'],
|
|
|
|
['hibernate', 'ENABLE_HIBERNATE'],
|
|
|
|
['environment-d', 'ENABLE_ENVIRONMENT_D'],
|
|
|
|
['binfmt', 'ENABLE_BINFMT'],
|
|
|
|
['coredump', 'ENABLE_COREDUMP'],
|
|
|
|
['resolve', 'ENABLE_RESOLVED'],
|
|
|
|
['logind', 'ENABLE_LOGIND'],
|
|
|
|
['hostnamed', 'ENABLE_HOSTNAMED'],
|
|
|
|
['localed', 'ENABLE_LOCALED'],
|
|
|
|
['machined', 'ENABLE_MACHINED'],
|
|
|
|
['networkd', 'ENABLE_NETWORKD'],
|
|
|
|
['timedated', 'ENABLE_TIMEDATED'],
|
|
|
|
['timesyncd', 'ENABLE_TIMESYNCD'],
|
|
|
|
['myhostname', 'HAVE_MYHOSTNAME'],
|
|
|
|
['firstboot', 'ENABLE_FIRSTBOOT'],
|
|
|
|
['randomseed', 'ENABLE_RANDOMSEED'],
|
|
|
|
['backlight', 'ENABLE_BACKLIGHT'],
|
|
|
|
['vconsole', 'ENABLE_VCONSOLE'],
|
|
|
|
['quotacheck', 'ENABLE_QUOTACHECK'],
|
|
|
|
['sysusers', 'ENABLE_SYSUSERS'],
|
|
|
|
['tmpfiles', 'ENABLE_TMPFILES'],
|
|
|
|
['hwdb', 'ENABLE_HWDB'],
|
|
|
|
['rfkill', 'ENABLE_RFKILL'],
|
|
|
|
['ldconfig', 'ENABLE_LDCONFIG'],
|
|
|
|
]
|
|
|
|
|
|
|
|
if get_option(pair[0])
|
|
|
|
conf.set(pair[1], 1)
|
|
|
|
endif
|
|
|
|
endforeach
|
|
|
|
|
2017-04-07 12:19:09 +08:00
|
|
|
want_tests = get_option('tests')
|
2017-04-08 13:55:38 +08:00
|
|
|
install_tests = get_option('install-tests')
|
2017-04-07 12:19:09 +08:00
|
|
|
tests = []
|
|
|
|
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
#####################################################################
|
|
|
|
|
|
|
|
if get_option('efi')
|
|
|
|
efi_arch = host_machine.cpu_family() # TODO: check this works at all
|
|
|
|
if efi_arch == 'ia32'
|
|
|
|
EFI_MACHINE_TYPE_NAME = 'ia32'
|
|
|
|
elif efi_arch == 'x86_64'
|
|
|
|
EFI_MACHINE_TYPE_NAME = 'x64'
|
|
|
|
elif efi_arch == 'aarch64'
|
|
|
|
EFI_MACHINE_TYPE_NAME = 'x64'
|
|
|
|
else
|
|
|
|
EFI_MACHINE_TYPE_NAME = efi_arch
|
|
|
|
endif
|
|
|
|
|
|
|
|
conf.set('ENABLE_EFI', 1)
|
|
|
|
conf.set_quoted('EFI_MACHINE_TYPE_NAME', EFI_MACHINE_TYPE_NAME)
|
|
|
|
endif
|
|
|
|
|
|
|
|
#####################################################################
|
|
|
|
|
|
|
|
config_h = configure_file(
|
|
|
|
output : 'config.h',
|
|
|
|
configuration : conf)
|
|
|
|
|
|
|
|
includes = include_directories('src/basic',
|
|
|
|
'src/shared',
|
|
|
|
'src/systemd',
|
|
|
|
'src/journal',
|
|
|
|
'src/resolve',
|
|
|
|
'src/timesync',
|
|
|
|
'src/login',
|
|
|
|
'src/udev',
|
|
|
|
'src/libudev',
|
|
|
|
'src/core',
|
|
|
|
'src/libsystemd/sd-bus',
|
|
|
|
'src/libsystemd/sd-device',
|
|
|
|
'src/libsystemd/sd-hwdb',
|
|
|
|
'src/libsystemd/sd-id128',
|
|
|
|
'src/libsystemd/sd-netlink',
|
|
|
|
'src/libsystemd/sd-network',
|
|
|
|
'src/libsystemd-network',
|
|
|
|
)
|
|
|
|
|
|
|
|
add_project_arguments('-include', 'config.h', language : 'c')
|
|
|
|
|
|
|
|
gcrypt_util_sources = files('src/shared/gcrypt-util.h',
|
|
|
|
'src/shared/gcrypt-util.c')
|
|
|
|
|
|
|
|
subdir('po')
|
|
|
|
subdir('catalog')
|
|
|
|
subdir('src/systemd')
|
|
|
|
subdir('src/basic')
|
|
|
|
subdir('src/libsystemd')
|
|
|
|
subdir('src/libsystemd-network')
|
|
|
|
subdir('src/journal')
|
|
|
|
subdir('src/login')
|
|
|
|
|
|
|
|
libjournal_core = static_library(
|
|
|
|
'journal-core',
|
|
|
|
libjournal_core_sources,
|
|
|
|
journald_gperf_c,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libbasic,
|
|
|
|
libsystemd_internal,
|
|
|
|
libsystemd_journal_internal],
|
|
|
|
install : false)
|
|
|
|
|
2017-04-11 02:13:40 +08:00
|
|
|
libsystemd_sym_path = '@0@/@1@'.format(meson.current_source_dir(), libsystemd_sym)
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
libsystemd = shared_library(
|
|
|
|
'systemd',
|
|
|
|
libsystemd_internal_sources,
|
2017-04-11 02:13:40 +08:00
|
|
|
libsystemd_journal_internal_sources,
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
version : '0.18.0',
|
|
|
|
include_directories : includes,
|
|
|
|
link_args : ['-shared',
|
2017-04-11 02:13:40 +08:00
|
|
|
'-Wl,--version-script=' + libsystemd_sym_path],
|
|
|
|
link_with : [libbasic],
|
|
|
|
dependencies : [threads,
|
|
|
|
librt,
|
|
|
|
libxz,
|
|
|
|
liblz4],
|
2017-04-10 01:11:49 +08:00
|
|
|
link_depends : libsystemd_sym,
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootlibdir)
|
|
|
|
|
|
|
|
############################################################
|
|
|
|
|
|
|
|
foreach tuple : [['myhostname', 'HAVE_MYHOSTNAME', []],
|
|
|
|
['systemd', '', []],
|
|
|
|
['mymachines', 'ENABLE_MACHINED', []],
|
|
|
|
['resolve', 'ENABLE_RESOLVED', [libdl]]]
|
|
|
|
|
|
|
|
condition = tuple[1] == '' or conf.get(tuple[1], 0) == 1
|
|
|
|
if condition
|
|
|
|
module = tuple[0]
|
|
|
|
extra_deps = tuple[2]
|
|
|
|
|
2017-04-10 01:11:49 +08:00
|
|
|
sym = 'src/nss-@0@/nss-@0@.sym'.format(module)
|
|
|
|
version_script_arg = join_paths(meson.current_source_dir(), sym)
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
|
|
|
|
shared_library(
|
|
|
|
'nss_' + module,
|
|
|
|
'src/nss-@0@/nss-@0@.c'.format(module),
|
|
|
|
version : '2',
|
|
|
|
include_directories : includes,
|
|
|
|
link_args : ['-shared',
|
2017-04-10 01:11:49 +08:00
|
|
|
'-Wl,--version-script=' + version_script_arg,
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
'-Wl,--undefined'],
|
|
|
|
link_with : [libsystemd_internal,
|
|
|
|
libbasic],
|
|
|
|
dependencies : [threads,
|
|
|
|
librt] + extra_deps,
|
2017-04-10 01:11:49 +08:00
|
|
|
link_depends : sym,
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootlibdir)
|
2017-04-11 00:26:39 +08:00
|
|
|
|
|
|
|
# We cannot use shared_module because it does not support version suffix.
|
|
|
|
# Unfortunately shared_library insists on creating the symlink…
|
|
|
|
meson.add_install_script('sh', '-c',
|
|
|
|
'rm $DESTDIR@0@/libnss_@1@.so'
|
|
|
|
.format(rootlibdir, module))
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
endif
|
|
|
|
endforeach
|
|
|
|
|
|
|
|
############################################################
|
|
|
|
|
|
|
|
subdir('src/libudev')
|
|
|
|
subdir('src/shared')
|
|
|
|
subdir('src/core')
|
|
|
|
subdir('src/udev')
|
|
|
|
subdir('src/network')
|
|
|
|
|
2017-04-07 12:19:09 +08:00
|
|
|
subdir('src/analyze')
|
|
|
|
subdir('src/journal-remote')
|
|
|
|
subdir('src/coredump')
|
|
|
|
subdir('src/hostname')
|
|
|
|
subdir('src/import')
|
|
|
|
subdir('src/kernel-install')
|
|
|
|
subdir('src/locale')
|
|
|
|
subdir('src/machine')
|
|
|
|
subdir('src/nspawn')
|
|
|
|
subdir('src/resolve')
|
|
|
|
subdir('src/timedate')
|
|
|
|
subdir('src/timesync')
|
|
|
|
subdir('src/vconsole')
|
2017-04-11 00:37:52 +08:00
|
|
|
subdir('src/sulogin-shell')
|
2017-04-07 12:19:09 +08:00
|
|
|
|
|
|
|
subdir('src/test')
|
|
|
|
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
executable('systemd',
|
|
|
|
systemd_sources,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libcore,
|
2017-04-11 05:11:10 +08:00
|
|
|
libshared,
|
|
|
|
libudev,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [threads,
|
|
|
|
librt,
|
|
|
|
libseccomp,
|
|
|
|
libselinux,
|
2017-04-10 02:08:53 +08:00
|
|
|
libmount,
|
|
|
|
libblkid],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-analyze',
|
|
|
|
systemd_analyze_sources,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libcore,
|
2017-04-11 05:11:10 +08:00
|
|
|
libudev,
|
|
|
|
libsystemd_internal,
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
libshared],
|
|
|
|
dependencies : [threads,
|
|
|
|
librt,
|
|
|
|
libseccomp,
|
|
|
|
libselinux,
|
2017-04-10 02:08:53 +08:00
|
|
|
libmount,
|
|
|
|
libblkid],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true)
|
|
|
|
|
|
|
|
executable('systemd-journald',
|
|
|
|
systemd_journald_sources,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libsystemd_journal_internal,
|
|
|
|
libjournal_core,
|
|
|
|
libshared,
|
|
|
|
libudev],
|
|
|
|
dependencies : [threads,
|
|
|
|
libxz,
|
|
|
|
liblz4],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-cat',
|
|
|
|
systemd_cat_sources,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libjournal_core,
|
|
|
|
libshared,
|
|
|
|
libudev],
|
|
|
|
dependencies : [threads],
|
|
|
|
install : true)
|
|
|
|
|
|
|
|
executable('journalctl',
|
|
|
|
journalctl_sources,
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libsystemd_journal_internal,
|
|
|
|
libshared,
|
|
|
|
libudev,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [threads,
|
|
|
|
libqrencode,
|
|
|
|
libxz,
|
|
|
|
liblz4],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootbindir)
|
|
|
|
|
|
|
|
executable('systemd-getty-generator',
|
|
|
|
'src/getty-generator/getty-generator.c',
|
|
|
|
install : true,
|
|
|
|
install_dir : systemgeneratordir,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared])
|
|
|
|
|
|
|
|
executable('systemd-debug-generator',
|
|
|
|
'src/debug-generator/debug-generator.c',
|
|
|
|
install : true,
|
|
|
|
install_dir : systemgeneratordir,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared])
|
|
|
|
|
|
|
|
executable('systemd-fstab-generator',
|
|
|
|
'src/fstab-generator/fstab-generator.c',
|
|
|
|
'src/core/mount-setup.c',
|
|
|
|
install : true,
|
|
|
|
install_dir : systemgeneratordir,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared])
|
|
|
|
|
|
|
|
if conf.get('ENABLE_ENVIRONMENT_D', 0) == 1
|
|
|
|
executable('30-systemd-environment-d-generator',
|
|
|
|
'src/environment-d-generator/environment-d-generator.c',
|
|
|
|
install : true,
|
|
|
|
install_dir : userenvgeneratordir,
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal])
|
2017-04-10 11:55:50 +08:00
|
|
|
|
|
|
|
meson.add_install_script(meson_make_symlink,
|
|
|
|
sysconfdir + '/environment',
|
|
|
|
environmentdir + '/99-environment.conf')
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_HIBERNATE', 0) == 1
|
|
|
|
executable('systemd-hibernate-resume-generator',
|
|
|
|
'src/hibernate-resume/hibernate-resume-generator.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : systemgeneratordir)
|
|
|
|
|
|
|
|
executable('systemd-hibernate-resume',
|
|
|
|
'src/hibernate-resume/hibernate-resume.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('HAVE_BLKID', 0) == 1
|
|
|
|
executable('systemd-gpt-auto-generator',
|
|
|
|
'src/gpt-auto-generator/gpt-auto-generator.c',
|
|
|
|
'src/basic/blkid-util.h',
|
|
|
|
install : true,
|
|
|
|
install_dir : systemgeneratordir,
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libsystemd_internal,
|
|
|
|
libshared,
|
|
|
|
libudev],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : libblkid)
|
|
|
|
|
|
|
|
executable('systemd-dissect',
|
|
|
|
'src/dissect/dissect.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_RESOLVED', 0) == 1
|
|
|
|
executable('systemd-resolved',
|
|
|
|
systemd_resolved_sources,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared,
|
2017-04-11 05:11:10 +08:00
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [threads,
|
|
|
|
libm,
|
|
|
|
libidn],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-resolve',
|
|
|
|
systemd_resolve_sources,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared,
|
2017-04-11 05:11:10 +08:00
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [threads,
|
|
|
|
libm,
|
|
|
|
libidn],
|
|
|
|
install : true)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_LOGIND', 0) == 1
|
|
|
|
executable('systemd-logind',
|
|
|
|
systemd_logind_sources,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [liblogind_core,
|
2017-04-11 05:11:10 +08:00
|
|
|
libudev,
|
|
|
|
libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [threads,
|
|
|
|
libacl],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('loginctl',
|
|
|
|
loginctl_sources,
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libudev,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [threads,
|
|
|
|
liblz4,
|
|
|
|
libxz],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootbindir)
|
|
|
|
|
|
|
|
executable('systemd-inhibit',
|
|
|
|
'src/login/inhibit.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootbindir)
|
|
|
|
|
|
|
|
if conf.get('HAVE_PAM', 0) == 1
|
2017-04-10 01:11:49 +08:00
|
|
|
version_script_arg = join_paths(meson.current_source_dir(), pam_systemd_sym)
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
shared_library(
|
|
|
|
'pam_systemd',
|
|
|
|
pam_systemd_c,
|
|
|
|
name_prefix : '',
|
|
|
|
include_directories : includes,
|
|
|
|
link_args : ['-shared',
|
2017-04-10 01:11:49 +08:00
|
|
|
'-Wl,--version-script=' + version_script_arg],
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libsystemd_internal,
|
|
|
|
libshared_static],
|
|
|
|
dependencies : [threads,
|
|
|
|
libpam,
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
libpam_misc],
|
2017-04-10 01:11:49 +08:00
|
|
|
link_depends : pam_systemd_sym,
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : pamlibdir)
|
|
|
|
endif
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('HAVE_PAM', 0) == 1
|
|
|
|
executable('systemd-user-sessions',
|
|
|
|
'src/user-sessions/user-sessions.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_EFI', 0) == 1
|
|
|
|
executable('bootctl',
|
|
|
|
'src/boot/bootctl.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
dependencies : [libblkid],
|
|
|
|
install : true)
|
|
|
|
endif
|
|
|
|
|
|
|
|
executable('systemd-socket-activate', 'src/activate/activate.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
dependencies : [threads],
|
|
|
|
install : true)
|
|
|
|
|
|
|
|
executable('systemctl', 'src/systemctl/systemctl.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [threads,
|
|
|
|
libcap,
|
|
|
|
libselinux,
|
|
|
|
libxz,
|
|
|
|
liblz4],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootbindir)
|
|
|
|
|
|
|
|
if conf.get('ENABLE_BACKLIGHT', 0) == 1
|
|
|
|
executable('systemd-backlight',
|
|
|
|
'src/backlight/backlight.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libudev],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_RFKILL', 0) == 1
|
|
|
|
executable('systemd-rfkill',
|
|
|
|
'src/rfkill/rfkill.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libudev],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
executable('systemd-system-update-generator',
|
|
|
|
'src/system-update-generator/system-update-generator.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : systemgeneratordir)
|
|
|
|
|
|
|
|
if conf.get('HAVE_LIBCRYPTSETUP', 0) == 1
|
|
|
|
executable('systemd-cryptsetup',
|
|
|
|
'src/cryptsetup/cryptsetup.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [libcryptsetup],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-cryptsetup-generator',
|
|
|
|
'src/cryptsetup/cryptsetup-generator.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
dependencies : [libcryptsetup],
|
|
|
|
install : true,
|
|
|
|
install_dir : systemgeneratordir)
|
|
|
|
|
|
|
|
executable('systemd-veritysetup',
|
|
|
|
'src/veritysetup/veritysetup.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
dependencies : [libcryptsetup],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-veritysetup-generator',
|
|
|
|
'src/veritysetup/veritysetup-generator.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [libcryptsetup],
|
|
|
|
install : true,
|
|
|
|
install_dir : systemgeneratordir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('HAVE_SYSV_COMPAT', 0) == 1
|
|
|
|
executable('systemd-sysv-generator',
|
|
|
|
'src/sysv-generator/sysv-generator.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : systemgeneratordir)
|
|
|
|
|
|
|
|
executable('systemd-rc-local-generator',
|
|
|
|
'src/rc-local-generator/rc-local-generator.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : systemgeneratordir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_HOSTNAMED', 0) == 1
|
|
|
|
executable('systemd-hostnamed',
|
|
|
|
'src/hostname/hostnamed.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('hostnamectl',
|
|
|
|
'src/hostname/hostnamectl.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_LOCALED', 0) == 1
|
|
|
|
executable('systemd-localed',
|
|
|
|
systemd_localed_sources,
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [libdl],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('localectl',
|
|
|
|
localectl_sources,
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_TIMEDATED', 0) == 1
|
|
|
|
executable('systemd-timedated',
|
|
|
|
'src/timedate/timedated.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('timedatectl',
|
|
|
|
'src/timedate/timedatectl.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_TIMESYNCD', 0) == 1
|
|
|
|
executable('systemd-timesyncd',
|
|
|
|
systemd_timesyncd_sources,
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [threads,
|
|
|
|
libm],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_MACHINED', 0) == 1
|
|
|
|
executable('systemd-machined',
|
|
|
|
systemd_machined_sources,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libmachine_core,
|
2017-04-11 05:11:10 +08:00
|
|
|
libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('machinectl',
|
|
|
|
'src/machine/machinectl.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [threads,
|
|
|
|
libxz,
|
|
|
|
liblz4],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootbindir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_IMPORTD', 0) == 1
|
|
|
|
executable('systemd-importd',
|
|
|
|
systemd_importd_sources,
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libsystemd_internal,
|
|
|
|
libshared],
|
|
|
|
dependencies : [threads],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-pull',
|
|
|
|
systemd_pull_sources,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
dependencies : [libcurl,
|
|
|
|
libz,
|
|
|
|
libbzip2,
|
|
|
|
libxz,
|
|
|
|
libgcrypt],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-import',
|
|
|
|
systemd_import_sources,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
dependencies : [libcurl,
|
|
|
|
libz,
|
|
|
|
libbzip2,
|
|
|
|
libxz],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-export',
|
|
|
|
systemd_export_sources,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
dependencies : [libcurl,
|
|
|
|
libz,
|
|
|
|
libbzip2,
|
|
|
|
libxz],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_REMOTE', 0) == 1 and conf.get('HAVE_LIBCURL', 0) == 1
|
|
|
|
executable('systemd-journal-upload',
|
|
|
|
systemd_journal_upload_sources,
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libsystemd_journal_internal,
|
|
|
|
libshared],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [threads,
|
|
|
|
libcurl,
|
|
|
|
libgnutls,
|
|
|
|
libxz,
|
|
|
|
liblz4],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_REMOTE', 0) == 1 and conf.get('HAVE_MICROHTTPD', 0) == 1
|
|
|
|
executable('systemd-journal-remote',
|
|
|
|
systemd_journal_remote_sources,
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libsystemd_journal_internal,
|
|
|
|
libshared],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [threads,
|
|
|
|
libmicrohttpd,
|
|
|
|
libgnutls,
|
|
|
|
libxz,
|
|
|
|
liblz4],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-journal-gatewayd',
|
|
|
|
systemd_journal_gatewayd_sources,
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libsystemd_journal_internal,
|
|
|
|
libshared],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [threads,
|
|
|
|
libmicrohttpd,
|
|
|
|
libgnutls,
|
|
|
|
libxz,
|
|
|
|
liblz4],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_COREDUMP', 0) == 1
|
|
|
|
executable('systemd-coredump',
|
|
|
|
systemd_coredump_sources,
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libsystemd_journal_internal,
|
|
|
|
libshared,
|
|
|
|
libsystemd],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [threads,
|
|
|
|
libacl,
|
|
|
|
libdw,
|
|
|
|
libxz,
|
|
|
|
liblz4],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('coredumpctl',
|
|
|
|
coredumpctl_sources,
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libsystemd_journal_internal,
|
|
|
|
libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [threads,
|
|
|
|
libxz,
|
|
|
|
liblz4],
|
|
|
|
install : true)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_BINFMT', 0) == 1
|
|
|
|
executable('systemd-binfmt',
|
|
|
|
'src/binfmt/binfmt.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
2017-04-10 11:55:05 +08:00
|
|
|
|
|
|
|
meson.add_install_script('sh', '-c',
|
|
|
|
mkdir_p.format(binfmtdir))
|
|
|
|
meson.add_install_script('sh', '-c',
|
|
|
|
mkdir_p.format(sysconfdir + '/binfmt.d'))
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_VCONSOLE', 0) == 1
|
|
|
|
executable('systemd-vconsole-setup',
|
|
|
|
'src/vconsole/vconsole-setup.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_RANDOMSEED', 0) == 1
|
|
|
|
executable('systemd-random-seed',
|
|
|
|
'src/random-seed/random-seed.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_FIRSTBOOT', 0) == 1
|
|
|
|
executable('systemd-firstboot',
|
|
|
|
'src/firstboot/firstboot.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
dependencies : [libcrypt],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootbindir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
executable('systemd-remount-fs',
|
|
|
|
'src/remount-fs/remount-fs.c',
|
|
|
|
'src/core/mount-setup.c',
|
|
|
|
'src/core/mount-setup.h',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-machine-id-setup',
|
|
|
|
'src/machine-id-setup/machine-id-setup-main.c',
|
|
|
|
'src/core/machine-id-setup.c',
|
|
|
|
'src/core/machine-id-setup.h',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootbindir)
|
|
|
|
|
|
|
|
executable('systemd-fsck',
|
|
|
|
'src/fsck/fsck.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-sleep',
|
|
|
|
'src/sleep/sleep.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-sysctl',
|
|
|
|
'src/sysctl/sysctl.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-ac-power',
|
|
|
|
'src/ac-power/ac-power.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-detect-virt',
|
|
|
|
'src/detect-virt/detect-virt.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true)
|
|
|
|
|
|
|
|
executable('systemd-delta',
|
|
|
|
'src/delta/delta.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true)
|
|
|
|
|
|
|
|
executable('systemd-escape',
|
|
|
|
'src/escape/escape.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootbindir)
|
|
|
|
|
|
|
|
executable('systemd-notify',
|
|
|
|
'src/notify/notify.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootbindir)
|
|
|
|
|
|
|
|
executable('systemd-volatile-root',
|
|
|
|
'src/volatile-root/volatile-root.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-cgroups-agent',
|
|
|
|
'src/cgroups-agent/cgroups-agent.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-path',
|
|
|
|
'src/path/path.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libsystemd_internal,
|
|
|
|
libshared],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true)
|
|
|
|
|
|
|
|
executable('systemd-ask-password',
|
|
|
|
'src/ask-password/ask-password.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootbindir)
|
|
|
|
|
|
|
|
executable('systemd-reply-password',
|
|
|
|
'src/reply-password/reply-password.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-tty-ask-password-agent',
|
|
|
|
'src/tty-ask-password-agent/tty-ask-password-agent.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootbindir)
|
|
|
|
|
|
|
|
executable('systemd-cgls',
|
|
|
|
'src/cgls/cgls.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true)
|
|
|
|
|
|
|
|
executable('systemd-cgtop',
|
|
|
|
'src/cgtop/cgtop.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true)
|
|
|
|
|
|
|
|
executable('systemd-initctl',
|
|
|
|
'src/initctl/initctl.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-mount',
|
|
|
|
'src/mount/mount-tool.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal,
|
|
|
|
libudev],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true)
|
|
|
|
|
2017-04-10 11:55:50 +08:00
|
|
|
meson.add_install_script(meson_make_symlink,
|
|
|
|
'systemd-mount', bindir + '/systemd-umount')
|
|
|
|
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
executable('systemd-run',
|
|
|
|
'src/run/run.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true)
|
|
|
|
|
|
|
|
executable('systemd-stdio-bridge',
|
|
|
|
'src/stdio-bridge/stdio-bridge.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true)
|
|
|
|
|
|
|
|
executable('busctl',
|
|
|
|
'src/busctl/busctl.c',
|
|
|
|
'src/busctl/busctl-introspect.c',
|
|
|
|
'src/busctl/busctl-introspect.h',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true)
|
|
|
|
|
|
|
|
if conf.get('ENABLE_SYSUSERS', 0) == 1
|
|
|
|
executable('systemd-sysusers',
|
|
|
|
'src/sysusers/sysusers.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootbindir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_TMPFILES', 0) == 1
|
|
|
|
executable('systemd-tmpfiles',
|
|
|
|
'src/tmpfiles/tmpfiles.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
dependencies : [libacl],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootbindir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_HWDB', 0) == 1
|
|
|
|
executable('systemd-hwdb',
|
|
|
|
'src/hwdb/hwdb.c',
|
|
|
|
'src/libsystemd/sd-hwdb/hwdb-internal.h',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootbindir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
if conf.get('ENABLE_QUOTACHECK', 0) == 1
|
|
|
|
executable('systemd-quotacheck',
|
|
|
|
'src/quotacheck/quotacheck.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
endif
|
|
|
|
|
|
|
|
executable('systemd-socket-proxyd',
|
|
|
|
'src/socket-proxy/socket-proxyd.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared,
|
2017-04-11 05:11:10 +08:00
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [threads],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-udevd',
|
|
|
|
systemd_udevd_sources,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libudev_core,
|
|
|
|
libudev_internal,
|
|
|
|
libsystemd_network,
|
2017-04-11 05:11:10 +08:00
|
|
|
libsystemd_internal,
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
libshared],
|
|
|
|
dependencies : [libkmod,
|
|
|
|
libidn,
|
|
|
|
libacl],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('udevadm',
|
|
|
|
udevadm_sources,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libudev_core,
|
|
|
|
libudev_internal,
|
|
|
|
libsystemd_network,
|
2017-04-11 05:11:10 +08:00
|
|
|
libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [libkmod,
|
|
|
|
libidn,
|
|
|
|
libacl],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootbindir)
|
|
|
|
|
|
|
|
executable('systemd-shutdown',
|
|
|
|
systemd_shutdown_sources,
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libudev],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-update-done',
|
|
|
|
'src/update-done/update-done.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-update-utmp',
|
|
|
|
'src/update-utmp/update-utmp.c',
|
|
|
|
include_directories : includes,
|
2017-04-11 05:11:10 +08:00
|
|
|
link_with : [libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [libaudit],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
if conf.get('HAVE_KMOD', 0) == 1
|
|
|
|
executable('systemd-modules-load',
|
|
|
|
'src/modules-load/modules-load.c',
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libshared],
|
|
|
|
dependencies : [libkmod],
|
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
2017-04-10 11:55:05 +08:00
|
|
|
|
|
|
|
meson.add_install_script('sh', '-c',
|
|
|
|
mkdir_p.format(modulesloaddir))
|
|
|
|
meson.add_install_script('sh', '-c',
|
|
|
|
mkdir_p.format(sysconfdir + '/modules-load.d'))
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
endif
|
|
|
|
|
|
|
|
executable('systemd-nspawn',
|
|
|
|
systemd_nspawn_sources,
|
|
|
|
'src/core/mount-setup.c', # FIXME: use a variable?
|
|
|
|
'src/core/mount-setup.h',
|
|
|
|
'src/core/loopback-setup.c',
|
|
|
|
'src/core/loopback-setup.h',
|
|
|
|
include_directories : [includes, include_directories('src/nspawn')],
|
|
|
|
link_with : [libfirewall,
|
2017-04-11 05:11:10 +08:00
|
|
|
libshared,
|
|
|
|
libudev,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
dependencies : [libacl,
|
|
|
|
libblkid,
|
|
|
|
libseccomp,
|
|
|
|
libselinux],
|
|
|
|
install : true)
|
|
|
|
|
|
|
|
executable('systemd-networkd',
|
|
|
|
systemd_networkd_sources,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libnetworkd_core,
|
|
|
|
libfirewall,
|
|
|
|
libsystemd_network,
|
|
|
|
libudev_internal,
|
2017-04-11 05:11:10 +08:00
|
|
|
libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('systemd-networkd-wait-online',
|
|
|
|
systemd_networkd_wait_online_sources,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libnetworkd_core,
|
2017-04-11 05:11:10 +08:00
|
|
|
libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootlibexecdir)
|
|
|
|
|
|
|
|
executable('networkctl',
|
|
|
|
networkctl_sources,
|
|
|
|
include_directories : includes,
|
|
|
|
link_with : [libsystemd_network,
|
2017-04-11 05:11:10 +08:00
|
|
|
libshared,
|
|
|
|
libsystemd_internal],
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
install : true,
|
|
|
|
install_dir : rootbindir)
|
|
|
|
|
2017-04-07 12:19:09 +08:00
|
|
|
############################################################
|
|
|
|
|
|
|
|
foreach tuple : tests
|
|
|
|
sources = tuple[0]
|
|
|
|
link_with = tuple[1].length() > 0 ? tuple[1] : [libshared]
|
|
|
|
dependencies = tuple[2]
|
|
|
|
condition = tuple.length() >= 4 ? tuple[3] : ''
|
|
|
|
type = tuple.length() >= 5 ? tuple[4] : ''
|
|
|
|
defs = tuple.length() >= 6 ? tuple[5] : []
|
|
|
|
incs = tuple.length() >= 7 ? tuple[6] : includes
|
|
|
|
|
|
|
|
name = sources[0].split('/')[-1].split('.')[0]
|
|
|
|
|
|
|
|
if condition == '' or conf.get(condition, 0) == 1
|
2017-04-08 13:55:38 +08:00
|
|
|
install = install_tests and type == ''
|
|
|
|
|
2017-04-07 12:19:09 +08:00
|
|
|
exe = executable(
|
|
|
|
name,
|
|
|
|
sources,
|
|
|
|
include_directories : incs,
|
|
|
|
link_with : link_with,
|
|
|
|
dependencies : dependencies,
|
2017-04-08 13:55:38 +08:00
|
|
|
c_args : defs,
|
|
|
|
install : install,
|
|
|
|
install_dir : testsdir)
|
|
|
|
|
2017-04-07 12:19:09 +08:00
|
|
|
if type == 'manual'
|
|
|
|
message('@0@ is a manual test'.format(name))
|
|
|
|
elif type == 'unsafe' and want_tests != 'unsafe'
|
|
|
|
message('@0@ is an unsafe test'.format(name))
|
|
|
|
else
|
|
|
|
test(name, exe, env : test_env)
|
|
|
|
endif
|
|
|
|
else
|
|
|
|
message('Not compiling @0@ because @1@ is not true'.format(name, condition))
|
|
|
|
endif
|
|
|
|
endforeach
|
|
|
|
|
2017-04-11 02:13:40 +08:00
|
|
|
test_libsystemd_sym = executable(
|
|
|
|
'test-libsystemd-sym',
|
|
|
|
test_libsystemd_sym_c,
|
|
|
|
include_directories : includes,
|
2017-04-11 05:35:55 +08:00
|
|
|
link_with : [libsystemd],
|
2017-04-11 02:13:40 +08:00
|
|
|
install : install_tests,
|
|
|
|
install_dir : testsdir)
|
|
|
|
test('test-libsystemd-sym',
|
|
|
|
test_libsystemd_sym)
|
|
|
|
|
2017-04-11 03:20:42 +08:00
|
|
|
test_libudev_sym = executable(
|
|
|
|
'test-libudev-sym',
|
|
|
|
test_libudev_sym_c,
|
|
|
|
include_directories : includes,
|
|
|
|
c_args : ['-Wno-deprecated-declarations'],
|
2017-04-11 05:35:55 +08:00
|
|
|
link_with : [libudev],
|
2017-04-11 03:20:42 +08:00
|
|
|
install : install_tests,
|
|
|
|
install_dir : testsdir)
|
|
|
|
test('test-libudev-sym',
|
|
|
|
test_libudev_sym)
|
|
|
|
|
2017-04-07 12:19:09 +08:00
|
|
|
############################################################
|
meson: build systemd using meson
It's crucial that we can build systemd using VS2010!
... er, wait, no, that's not the official reason. We need to shed old systems
by requring python 3! Oh, no, it's something else. Maybe we need to throw out
345 years of knowlege accumulated in autotools? Whatever, this new thing is
cool and shiny, let's use it.
This is not complete, I'm throwing it out here for your amusement and critique.
- rules for sd-boot are missing. Those might be quite complicated.
- rules for tests are missing too. Those are probably quite simple and
repetitive, but there's lots of them.
- it's likely that I didn't get all the conditions right, I only tested "full"
compilation where most deps are provided and nothing is disabled.
- busname.target and all .busname units are skipped on purpose.
Otherwise, installation into $DESTDIR has the same list of files and the
autoconf install, except for .la files.
It'd be great if people had a careful look at all the library linking options.
I added stuff until things compiled, and in the end there's much less linking
then in the old system. But it seems that there's still a lot of unnecessary
deps.
meson has a `shared_module` statement, which sounds like something appropriate
for our nss and pam modules. Unfortunately, I couldn't get it to work. For the
nss modules, we need an .so version of '2', but `shared_module` disallows the
version argument. For the pam module, it also didn't work, I forgot the reason.
The handling of .m4 and .in and .m4.in files is rather awkward. It's likely
that this could be simplified. If make support is ever dropped, I think it'd
make sense to switch to a different templating system so that two different
languages and not required, which would make everything simpler yet.
v2:
- use get_pkgconfig_variable
- use sh not bash
- use add_project_arguments
v3:
- drop required:true and fix progs/prog typo
v4:
- use find_library('bz2')
- add TTY_GID definition
- define __SANE_USERSPACE_TYPES__
- use join_paths(prefix, ...) is used on all paths to make them all absolute
v5:
- replace all declare_dependency's with []
- add more conf.get guards around optional components
v6:
- drop -pipe, -Wall which are the default in meson
- use compiler.has_function() and compiler.has_header_symbol instead of the
hand-rolled checks.
- fix duplication in 'liblibsystemd' library name
- use the right .sym file for pam_systemd
- rename 'compiler' to 'cc': shorter, and more idiomatic.
v7:
- use ENABLE_ENVIRONMENT_D not HAVE_ENVIRONMENT_D
- rename prefix to prefixdir, rootprefix to rootprefixdir
("prefix" is too common of a name and too easy to overwrite by mistake)
- wrap more stuff with conf.get('ENABLE...') == 1
- use rootprefix=='/' and rootbindir as install_dir, to fix paths under
split-usr==true.
v8:
- use .split() also for src/coredump. Now everything is consistent ;)
- add rootlibdir option and use it on the libraries that require it
v9:
- indentation
v10:
- fix check for qrencode and libaudit
v11:
- unify handling of executable paths, provide options for all progs
This makes the meson build behave slightly differently than the
autoconf-based one, because we always first try to find the executable in the
filesystem, and fall back to the default. I think different handling of
loadkeys, setfont, and telinit was just a historical accident.
In addition to checking in $PATH, also check /usr/sbin/, /sbin for programs.
In Fedora $PATH includes /usr/sbin, (and /sbin is is a symlink to /usr/sbin),
but in Debian, those directories are not included in the path.
C.f. https://github.com/mesonbuild/meson/issues/1576.
- call all the options 'xxx-path' for clarity.
- sort man/rules/meson.build properly so it's stable
2017-04-05 11:03:47 +08:00
|
|
|
|
|
|
|
make_directive_index_py = find_program('tools/make-directive-index.py')
|
|
|
|
make_man_index_py = find_program('tools/make-man-index.py')
|
|
|
|
|
|
|
|
subdir('units')
|
|
|
|
subdir('sysctl.d')
|
|
|
|
subdir('sysusers.d')
|
|
|
|
subdir('tmpfiles.d')
|
|
|
|
subdir('rules')
|
|
|
|
subdir('hwdb')
|
|
|
|
subdir('network')
|
|
|
|
subdir('man')
|
|
|
|
subdir('shell-completion/bash')
|
|
|
|
subdir('shell-completion/zsh')
|
|
|
|
subdir('docs/sysvinit')
|
|
|
|
subdir('docs/var-log')
|
|
|
|
|
|
|
|
# FIXME: figure out if the warning is true:
|
|
|
|
# https://github.com/mesonbuild/meson/wiki/Reference-manual#install_subdir
|
|
|
|
install_subdir('factory/etc',
|
|
|
|
install_dir : factorydir)
|
|
|
|
|
|
|
|
|
|
|
|
install_data('xorg/50-systemd-user.sh',
|
|
|
|
install_dir : xinitrcdir)
|
|
|
|
install_data('system-preset/90-systemd.preset',
|
|
|
|
install_dir : systempresetdir)
|
|
|
|
install_data('README',
|
|
|
|
'NEWS',
|
|
|
|
'CODING_STYLE',
|
|
|
|
'DISTRO_PORTING',
|
|
|
|
'ENVIRONMENT.md',
|
|
|
|
'LICENSE.GPL2',
|
|
|
|
'LICENSE.LGPL2.1',
|
|
|
|
'src/libsystemd/sd-bus/GVARIANT-SERIALIZATION',
|
|
|
|
install_dir : docdir)
|
2017-04-07 02:33:27 +08:00
|
|
|
|
2017-04-10 11:55:05 +08:00
|
|
|
meson.add_install_script('sh', '-c', mkdir_p.format(systemdstatedir))
|
|
|
|
meson.add_install_script('sh', '-c', 'touch $DESTDIR@0@'.format(prefixdir))
|
|
|
|
|
2017-04-07 02:33:27 +08:00
|
|
|
############################################################
|
|
|
|
|
|
|
|
if git.found() and etags.found()
|
|
|
|
all_files = run_command(
|
|
|
|
git,
|
|
|
|
['--git-dir=@0@/.git'.format(meson.source_root()),
|
|
|
|
'ls-files',
|
|
|
|
':/*.[ch]'])
|
|
|
|
all_files = files(all_files.stdout().split())
|
|
|
|
|
|
|
|
custom_target(
|
|
|
|
'TAGS',
|
|
|
|
output : 'TAGS',
|
|
|
|
input : all_files,
|
|
|
|
command : [etags, '-o', '@OUTPUT@'] + all_files)
|
|
|
|
endif
|