Boost tests are disabled on Windows for now because the detection
is actually completely broken. Once that's fixed (after the release)
we can enable it again.
- Adds a `crate_type` kwarg to library targets, allowing the different
types of Rust [linkage][1].
- Shared libraries use the `dylib` crate type by default, but can also
be `cdylib`
- Static libraries use the `rlib` crate type by default, but can also
be `staticlib`
- If any Rust target has shared library dependencies, add the
appropriate linker arguments, including rpath for the sysroot of the
Rust compiler
[1]: https://doc.rust-lang.org/reference/linkage.html
It is not feasible to test all failure modes by creating projects in
`test cases/failing` that would be an explosion of files, and that
mechanism is too coarse anyway. We have no way to ensure that the
expected error is being raised.
See FailureTests.test_dependency for an example.
This is really useful when debugging test failures. Without a stack
trace, you have to grep the source code for the error message.
Also set this in run_tests.py.
This class now consolidates a lot of the logic that each external
dependency was duplicating in its class definition.
All external dependencies now set:
* self.version
* self.compile_args and self.link_args
* self.is_found (if found)
* self.sources
* etc
And the abstract ExternalDependency class defines the methods that
will fetch those properties. Some classes still override that for
various reasons, but those should also be migrated to properties as
far as possible.
Next step is to consolidate and standardize the way in which we call
'configuration binaries' such as sdl2-config, llvm-config, pkg-config,
etc. Currently each class has to duplicate code involved with that
even though the format is very similar.
Currently only pkg-config supports multiple version requirements, and
some classes don't even properly check the version requirement. That
will also become easier now.
This commit syncs the list with gnome-builder, which updates its list
in https://bugzilla.gnome.org/show_bug.cgi?id=782296.
llvm-config40 is added becaue LLVM 4.0 becomes a stable release.
llvm-config-5.0 is added to the bottom of the list because it is still
a development snapshot (svn trunk).
While g-ir-scanner's compatible -I and -D flags cover what most pkg-config
files use, there's no guarantee that files don't set anything more exotic
that conflicts with the tool's own options.
For a real world example, mozjs-38 has '-include some-header-file.h', which
translates to '--include nclude another-file-to-scan.h' for the scanner;
unless for some reason there's an 'nclude' GIR available on the system,
the target will thus fail.
Avoid this case by pointing g-ir-scanner to the correct pkg-config file
instead of passing any cflags directly.
When the CCompiler.links method call in CCompiler.find_library fails,
find_library resorts to finding the library file itself. In this second
case, the return value is not a list, whereas if links suceeds, the
return value is a list. Make it so that find_library returns a list
in either case.
This is more reliable, and more accurate. For instance, this means
arguments in commands aren't surrounded by `'` on Linux unless that
is actually needed by that specific argument.
There is no equivalent helper for Windows, so we keep the old
behaviour for that.
Now we aggressively de-dup the list of libraries used while linking,
and when linking with GNU ld we have to enclose all static libraries
with -Wl,--start-group and -Wl,--end-group to force the linker to
resolve all symbols recursively. This is needed when static libraries
have circular deps on each other (see included test).
The --start/end-group change is also needed for circular dependencies
between static libraries because we no longer recursively list out all
library dependencies.
The size of build.ninja for GStreamer is now down to 6.1M from 20M,
and yields a net reduction in configuration time of 10%
We were doing this on the basis of an old comment, but there was no
test for it and I couldn't reproduce the issue with clang on Linux
at all.
Let's add a (somewhat comprehensive) test and see if it breaks
anywhere if we stop doing this.
Halves the size of gstreamer's build.ninja from 20M to 8.7M
Closes https://github.com/mesonbuild/meson/issues/1057
This is useful when build_machine appears to be compatible with
host_machine, but actually isn't. For example when:
- build_machine is macOS and host_machine is the iOS Simulator
- the build_machine's libc is glibc but the host_machine libc is uClibc
- code relies on kernel features not available on the build_machine
There are cases where we need to specify arguments to gtkdoc-mkdb, like
telling it to scan extensions that are not '.h' and '.c'. Let's add a
new named argument to gnome.gtkdoc(), as well as the plumbing needed for
the gtk-doc helper script.
This was already mostly working before due to how the linker arguments were constructed, but this will now be more resilient to the case where dynamic libraries only are present.
- supported as "Link with Libraries".
- Frameworks added as a group to the project
- no need to specify custom linker flags anymore, xcodeproj works just like ninja
example mason.build file :
dep_main += [dependency('appleframeworks', modules : ['Foundation', 'AppKit', 'IOKIT', 'QuartzCore', 'OpenGL', 'GLUT', 'OpenAL'], required : true)]
Nirbheek Chauhan
Patrick Griffis
Adam C. Foltzer
Bug
Jan Chren (rindeal)
Ting-Wei Lan
Alan
Jussi Pakkanen
Florian Müllner
Paulo Antonio Alvarez
Rene Lopez
Bálint Aradi
Ole André Vadla Ravnås
Guillaume Poirier-Morency
Alistair Thomas
Emmanuele Bassi
Goncalo Carvalho
Rob Doolittle
blackbox