has_function() tries to link an example program using the function
to see if it is available, but with clang on 64bit Windows this
example always already failed at the compile step:
error: cast from pointer to smaller type 'long' loses information
long b = (long) a;
This is due to long!=pointer with LLP64
Change from "long" to "long long" which is min 64bit and should always
fit a pointer. While "long long" is strictly a C99 feature every
non super ancient compiler should still support it.
Currently deprecation features use the same logic as new features, but
that doesn't work correctly. FeatureNew wants to warn about cases where
you claim to support >= 0.40, but use a feature from 0.42; deprecation
wants to warn when you claim to support >= 0.50, but use a feature that
was replaced in 0.45.
To make this work we need to invert the version check in the deprecation
function, so that if the deprecation is 0.45, and the supported version
is >= 0.50, we get a true not a false.
Currently The Deprecated and New features checkers share an attribute
through a base class that should be per class. We need to duplicate this
and move it into each of the sublcasses
Fixes#7080
The system tool is always the wrong thing to use and cause hard to debug
issues when trying to link system libraries with cross built binaries.
The ExternalDependency base class already had a method to deal with
this, used by PkgConfigDependency and QtBaseDependency, so it should
make things more consistent.
The builtin check had a special case that if a header was provided and
the function wasn't defined, it would ignore the builtin to avoid
non-functional builtins (for example __builtin_posix_memalign in MSYS2).
GCC 10 gained support for __has_builtin() which now skipps this check
and because __has_builtin(__builtin_posix_memalign) returns true the
non functional builtin is now reported as available.
To get the old behaviour back move the special case in front of the actual
availability check.
Fixes#7113
As stated by the doc, default to install python sources to purelib
location, as they should not depend on platform.
This also fixes discrepancy between get_install_dir() and
install_sources() locations.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Discussions in #6524 have shown that there are various possible uses of the
kconfig module and even disagreements in the exact file format between
Python-based kconfiglib and the tools in Linux. Instead of trying to
reconcile them, just rename the module to something less suggestive and
leave any policy to meson.build files.
In the future it may be possible to add some kind of parsing through
keyword arguments such as bool_true, quoted_strings, etc. and possibly
creation of key-value lists too. For now, configuration_data objects
provide an easy way to access quoted strings. Note that Kconfig stores
false as "absent" so it was already necessary to write "x.has_key('abc')"
rather than the more compact "x['abc']". Therefore, having to use
configuration_data does not make things much more verbose.
The comment for this code is correct, but the code itself isn't. The way
it's implemented in a cross compile we don't look at PKG_CONFIG_PATH at
all.
Fixes: #7062
Gtest can output junit results with a command line switch. We can parse
this to get more detailed results than the returncode, and put those in
our own Junit output. We basically just throw away the top level
'testsuites' object, then fixup the names of the tests, and shove that
into our junit.
Cosmetic tweak to the error message for incdir() with an absolute path.
Don't split the message in the middle of a sentence at a point which may
or may not correspond to the terminal width.
For the sake of a consistent error message (irrespective of if 'valac' is
present or not), check if the 'c' language is present if we are adding
'vala' before (rather than after) we do compiler detection.
Generally, we'd want to use str() rather than repr() in error messages
anyhow, as that explicitly gives something designed to be read by
humans.
Sometimes {!r} is being used as a shortcut to avoid writing the quotes
in '{!s}'.
Unfortunately, these things aren't quite the same, as the repr of a
string containing '\' (the path separator on Windows) will have those
escaped.
We don't have a good string representation to use for the arbitrary
internal object used as an argument for install_data() when it's neither
a string nor file (which doesn't lead to a good error message), so drop
that for the moment.
Currently, colourize_console is a constant, set at process
initialization.
To allow the actual stdout to be easily compared with the expected when
running tests, we want to allow colourization to be on for the test
driver, but not for the in-process configure done by run_configure,
which has stdout redirected from a tty to a pipe.
v2:
Cache _colorize_console per file object
v3:
Reset cache on setup_console()
If the feature hadn't been broken in the first place it would have
worked on them anyway, so we might as well expose it. I'm loathe to do
it because one of the best features of meson in a mixed C/C++ code base
is that meson figures out the right linker every time, but there are
cases people have where they want to force a linker. We'll let them keep
the pieces.
Currently it does nothing, as the field is read too late, and additional
languages have already been considered. As such if the language
requested is closer to C (for example you want C but have a C++ source
with only extern C functions) then link_langauge is ignored.
Fixes#6453
This adds support for Files, CustomTarget, Indexs of CustomTargets,
ConfigureFiles, ExternalPrograms, and Executables.
Fixes: #1234Fixes: #3552Fixes: #6175
When wiping a build tree with --wipe, every entry in the build directory
is removed with mesonlib.windows_proof_rmtree() for directories and
mesonlib.windows_proof_rm() for other files. Symlinks to directories are
considered directories, resulting in the former being called. This
causes an exception to be raised, as the implementation calls
shutil.rmtree(), which isn't allowed on symlinks.
Fix this by using mesonlib.windows_proof_rm() for symlinks.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
For instance if C:/Program Files (x86)/folder is passed to _guess_files, it would resolve to ['C:/Program Files', '(x86)/folder'] since C:/Program Files is an actual file location that can exist.
* dependency: log cached or skipped dependencies with reference to modules
If the dependency is a special dependency which takes modules, the
modules get cached separately and probably reference different
pkg-config files. It's very plausible that we have multiple
dependencies, using different modules. For example:
Run-time dependency qt5 (modules: Core) found: YES 5.14.2 (pkg-config)
Dependency qt5 skipped: feature gui disabled
Obviously this makes no sense, because of course we found qt5 and even
used it. The second line is a lot more readable if it shows this:
Dependency qt5 (modules: Widgets) skipped: feature gui disabled
Similar confusion abounds in the case where a module is found in the
cache -- which module, exactly, has been found here:
Dependency qt5 found: YES 5.14.2 (cached)
Rewrite the dependency function to *consistently* pass around (and use!)
the display_name even in cases where we know it isn't anonymous (this is
just more correct anyway), and make it serve a dual purpose by also
appending the list of modules just like we do for pretty-printing that a
dependency has just been found for the first time.
* fixup! dependency: log cached or skipped dependencies with reference to modules
pointlessly cast modules to str, as they cannot be anything else. But we
want to fail later on, with something more friendly than a stacktrace.
boost/wx have special exceptions for people passing an integer there.
- ExternalProgramHolder has path() method while CustomTargetHolder and
BuildTargetHolder have full_path().
- The returned ExternalProgramHolder's path() method was broken, because
build.Executable object has no get_path() method, it needs the
backend.
- find_program('overridden_prog', version : '>=1.0') was broken because
it needs to execute the exe that is not yet built. Now assume the
program has the (sub)project version.
- If the version check fails, interpreter uses
ExternalProgramHolder.get_name() for the error message but
build.Executable does not implement get_name() method.
Eric Lemanissier
Ole André Vadla Ravnås
Christoph Reiter
Daniel Mensinger
Dylan Baker
Xavier Claessens
Jussi Pakkanen
Marc-André Lureau
Andrei Alexeyev
Szabi Tolnai
Reza Housseini
Paolo Bonzini
James Hilliard
Eric Engestrom
GustavoLCR
Michael Brockus
Jon Turney
Laurent Pinchart
Benjamin Frye
georgev93
Eli Schwartz