Currently the Qt Dependencies still use the old "combined" method for
dependencies with multiple ways to be found. This is problematic as it
means that `get_variable()` and friends don't work, as the dependency
can't implement any of those methods. The correct solution is to make
use of multiple Dependency instances, and a factory to tie them
together. This does that.
To handle QMake, I've leveraged the existing config-tool mechanism,
which allows us to save a good deal of code, and use well tested code
instead of rolling more of our own code.
The one thing this doesn't do, but we probably should, is expose the
macOS ExtraFrameworks directly, instead of forcing them to be found
through QMake. That is a problem for another series, and someone who
cares more about macOS than I do.
It's a method on the QtDependeny that exists purely for the consumption
of the qt module (in the form, return some stuff the module makes into
an instance variable). So put it where it actually belongs, and pass the
qt dependency into it.
Dependencies is already a large and complicated package without adding
programs to the list. This also allows us to untangle a bit of spaghetti
that we have.
All changes were created by running
"pyupgrade --py3-only --keep-percent-format"
and committing the results. I have not touched string formatting for
now.
- use set literals
- simplify .format() parameter naming
- remove __future__
- remove default "r" mode for open()
- use OSError rather than compatibility aliases
- remove stray parentheses in function(generator) scopes
- ModuleState is now a real class that will have methods in the future
for actions modules needs, instead of using interpreter internal API.
- New ModuleObject base class, similar to InterpreterObject, that should
be used by all objects returned by modules. Its methods gets the
ModuleState passed as first argument. It has a `methods` dictionary to
define what is public API that can be called from build definition.
- Method return value is not required to be a ModuleReturnValue any
more, it can be any type that interpreter can holderify, including
ModuleObject.
- Legacy module API is maintained until we port all modules.
In the future modules should be updated:
- Use methods dict.
- Remove snippets.
- Custom objects returned by modules should all be subclass of
ModuleObject to get the state iface in their methods.
- Modules should never call into interpreter directly and instead state
object should have wrapper API.
- Stop using ModuleReturnValue in methods that just return simple
objects like strings. Possibly remove ModuleReturnValue completely
since all objects that needs to be processed by interpreter (e.g.
CustomTarget) should be created through ModuleState API.
Rather than having to manually build the locale aware man paths with
`install_data('foo.fr.1', install_dir: join_paths(get_option('mandir'), 'fr', 'man1'), rename: 'foo.1')`
Support doing
`install_man('foo.fr.1', locale: 'fr')`
By default all subprojects are installed. If --skip-subprojects is given
with no value only the main project is installed. If --skip-subprojects
is given with a value, it should be a coma separated list of subprojects
to skip and all others will be installed.
Fixes: #2550.
This has a couple of advantages over rolling it by hand:
1. it correctly handles include_directories objects, which is always
handy
2. it correctly generates a depfile for you, which makes it more
reliable
3. it requires less typing
Re-implement it in backend using the same code path as for
custom_target(). This for example handle setting PATH on Windows when
command is an executable.
Various GNOME projects have scripts that does similar task, better do it
directly in meson. This ensures it's done correctly regarding usage of
subprojects and pkg-config. See for example this gtk bug:
https://gitlab.gnome.org/GNOME/gtk/-/issues/3626.
Fixes: #8268
Following #7890, this patch introduces the ability to read the contents
of a file to the fs module.
This patch introduces the ability to read files at configure time, but
has some restrictions:
- binary files are not supported (I don't think this will prove a
problem, and if people are wanting to do something with binary
files, they should probably be shelling out to their own script).
- Only files outside the build directory allowed. This limitation
should prevent build loops.
Given that reading an arbitrary file at configure time can affect the
configuration in almost arbitrary ways, meson should force a reconfigure
when the given file changes. This is non-configurable, but this can
easily be changed with a future keyword argument.
On Windows this would fail because of missing DLL:
```
mylib = library(...)
exe = executable(..., link_with: mylib)
meson.add_install_script(exe)
```
The reason is on Windows we cannot rely on rpath to find libraries from
build directory, they are searched in $PATH. We already have all that
mechanism in place for custom_target() using ExecutableSerialisation
class, so reuse it for install/dist/postconf scripts too.
This has bonus side effect to also use exe_wrapper for those scripts.
Fixes: #8187
we have two functions to do the exact same thing, and they're basically
implemented the same way. Instead, let's just use the BuildTarget one,
as it's more generally available.
Currently InstallDir is part of the interpreter, and is an Interpreter
object, which is then put in the Build object. This is a layering
violation, the interperter should have a Holder for build data. This
patch fixes that.
This is PEP8 convention for a const variable. Also, make the type
Mapping, which doesn't have mutation methods. This means mypy will warn
us if someone tries to change this.
Like other language specific modules this module is module for holding
rust specific helpers. This commit adds a test() function, which
simplifies using rust's internal unittest mechanism.
Rust tests are generally placed in the same code files as they are
testing, in contrast to languages like C/C++ and python which generally
place the tests in separate translation units. For meson this is
somewhat problematic from a repetition point of view, as the only
changes are generally adding --test, and possibly some dependencies.
The rustmod.test() method provides a mechanism to remove the repatition:
it takes a rust target, copies it, and then addes the `--test` option,
then creates a Test() target with the `rust` protocol. You can pass
additional dependencies via the `dependencies` keyword. This all makes
for a nice, DRY, test definition.
This patches takes the options work to it's logical conclusion: A single
flat dictionary of OptionKey: UserOptions. This allows us to simplify a
large number of cases, as we don't need to check if an option is in this
dict or that one (or any of 5 or 6, actually).
I would have prefered to do these seperatately, but they are combined in
some cases, so it was much easier to convert them together.
this eliminates the builtins_per_machine dict, as it's duplicated with
the OptionKey's machine parameter.
See https://gitlab.gnome.org/GNOME/glib/-/issues/600
`volatile` was previously mistakenly used in GLib to indicate that a
variable was accessed atomically or otherwise multi-threaded. It’s not
meant for that, and up to date compilers (like gcc-11) will rightly warn
about it.
Drop the `volatile` qualifiers.
Based on a patch by Jeff Law.
See also http://isvolatileusefulwiththreads.in/c/.
Signed-off-by: Philip Withnall <pwithnall@endlessos.org>
The order of elements in sets cannot be relied upon, because the hash
values are randomized by Python. Whenever sets are converted to lists
we need to keep their order stable, or random changes in the command line
cause ninja to rebuild a lot of files unnecessarily. To stabilize them,
use either sort or OrderedSet. Sorting is not always applicable, but it
can be faster because it's done in C and it can produce slightly nicer
output.
Also, remove the possibility of passing in a compiler instance to
min_driver_version. This is because the NVCC compiler instance is,
as of CUDA Toolkit 11.0, no longer guaranteed to be versioned
identically to the toolkit itself.
This fix a regression caused by
https://github.com/mesonbuild/meson/pull/7837, it wanted to make
InternalDependency's external_deps private but has side effect of making
StatisLibrary's external_deps private too.
It is technically correct to make them private, but Meson used to make
StaticLibrary deps public so they are usable without `pkg-config
--static` when we know there is only a static library available.
Fixes: #7929.
It's an easy mistake to do this:
gir1 = gnome.generate_gir(...)
gir2 = gnome.generate_gir(...
includes: ['GObject-2.0', gir1])
This fails with an error:
ERROR: Gir includes must be str, GirTarget, or list of them.
The issue is that the 'gir1[0]' should be passed instead of 'gir1'.
To make the problem slightly clearly, tweak the error message to be:
ERROR: Gir includes must be str, GirTarget, or list of them. Got TypelibTarget.
This fixes glib-2.0-uninstalled.pc file.
GLib does `extra_cflags : ['-I${libdir}/glib-2.0/include']` because some
of its headers gets installed there. But when used uninstalled that path
makes no sense and pkg-config aborts because ${libdir} is not defined.
This cannot be worked around by GLib because Meson does not allow
setting different `extra_cflags` for -uninstalled.pc, and does not allow
setting libdir in `uninstalled_variables`.
When glib is a subproject we should use glib-compile-resources it
overrides using find_program() in the case it is not installed on the
build machine. With old glib version we have to run
glib-compile-resources at configure time to generate the list of
dependencies, but not when glib is recent enough.
This fixes a regression introduced by
https://github.com/mesonbuild/meson/pull/7488.
InternalDependency's ext_deps previously where simply ignored, but that
PR has effect to add many more public Requires in generated pc files.
Dylan Baker
Xavier Claessens
Olexa Bilaniuk
Eli Schwartz
Leif Middelschulte
Jason Woodward
Luca Weiss
Luke Drummond
Antonin Décimo
Philip Withnall
Paolo Bonzini
Sam Thursfield