has_function(prefix: '...') is useless to check the difference between
builtins and external library functions. It has code to detect
"builtins" that misfires and reports that iconv_open is defined as a
builtin on mingw, but only if you include the header.
Instead compile an open-coded test file that this iconv dependency
implementation fully controls, that doesn't get up to imaginative edge
cases like trying to find `__builtin_iconv_open`.
Fixes commit db1fa702f3, which merely
moved the brokenness over one step to the right (by breaking mingw
instead of freebsd)
Fixes https://github.com/mesonbuild/meson/pull/9632#issuecomment-979581509
This header is required anyway. And the compile test for linking to libc
with the iconv_open symbol, can succeed when we try to use the literal
symbol name without includes, but fail later during project build,
because actually including iconv.h might redefine the function to match
a forked symbol. This happens when GNU iconv is installed as a
standalone library on systems that have a less fully-featured iconv
implementation.
So, by including the header in has_function, we ensure that we test
against the default resolved header. In the event that the symbol which
is #define'd by the header is 'libiconv_open', linking will fail against
libc even when a builtin iconv does exist, and we will fall back to the
iconv dependency that provides -liconv (and which is needed to properly
use the default header).
Fixes compiling against the iconv dependency on e.g. FreeBSD when the
libiconv port is installed.
Contrary to most system method checks, zlib currently functions as a
whitelist of OSes. This isn't really needed however. The first special
case for OSes that provide zlib as part of the base OS is worth keeping.
However, the elif for windows is more than generic enough to allow any
other potential OSes to try. Just make it a simplie if/else instead.
Depending on whether hdf5 is compiled with parallel support, the
same config-tool program may be installed with a mysterious "p" in the
name. In this case, dependency lookup will totally fail, unless of
course you use the superior pkg-config interface in which case you get a
predictable name.
Work around this insanity by checking for both types of config-tool
name.
Fixes#9555
This is broken and terrible and thus completely unusable. Don't torture
users by finding pkg-config on Windows, thus permitting the pkg-config
lookup of several dependencies that do not actually work -- which then
fails at build time.
This also breaks CI for the wrapdb, because Strawberry Perl is provided
as part of the base image for the OS (yes, even though it is terribly
broken!!!) and anything that depends on e.g. zlib will "find" zlib
because of this broken disaster, even though it should use the wrapdb
subproject of zlib.
It is assumed no one actually wants to mix Strawberry Perl and meson. In
fact, some projects, such as gst-build, already unconditionally error
out if Strawberry Perl is detected in PATH:
error('You have Strawberry Perl in PATH which is known to cause build issues with gst-build. Please remove it from PATH or uninstall it.')
Other projects (postgresql) actually do want to build perl extensions,
and link to the perl dlls, but absolutely under no circumstances ever
want to use its pkg-config implementation. ;)
Let's solve this problem by just considering this to not be a valid
pkg-config, let the user find another or not have one at all.
This change "solves"
https://github.com/StrawberryPerl/Perl-Dist-Strawberry/issues/11
There are a bunch of cases in a single function where we would want to
log the detected path of pkg-config. Formatting this is awkward. Define
it once, then use f-strings everywhere. :D
CMakes `target_link_libraries()` supports certain keywords to
only enable specific libraries for specific CMake configurations.
We now try our best to replicate this for Meson dependencies.
Fixes#9197
'{}'.format('foo') for any given value of 'foo' (in this case, a
function returning a string), can always just be 'foo' directly, which
is a lot more readable.
All changes were created by running
"pyupgrade --py3-only"
and committing the results. Although this has been performed in the
past, newer versions of pyupgrade can automatically catch more
opportunities, notably list comprehensions can use generators instead,
in the following cases:
- unpacking into function arguments as function(*generator)
- unpacking into assignments of the form x, y = generator
- as the argument to some builtin functions such as min/max/sorted
Also catch a few creeping cases of new code added using older styles.
Dependencies are currently printed as
[<mesonbuild.mlog.AnsiDecorator object at 0x7faa85aeac70>, ' ', <mesonbuild.mlog.AnsiDecorator object at 0x7faa85aeab50>]
This was introduced in commit adb1b2f3f6, due to
an incorrect type annotation on the AnsiText constructor. Fix both the
annotation and the usage.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
We have a lot of these. Some of them are harmless, if unidiomatic, such
as `if (condition)`, others are potentially dangerous `assert(...)`, as
`assert(condtion)` works as expected, but `assert(condition, message)`
will result in an assertion that never triggers, as what you're actually
asserting is `bool(tuple[2])`, which will always be true.
This boolean parameter is added to check_and_set_roots() and detect_lib_dirs()
when true it will first search the compiler library directories before checking
the standard lib directories. This is set to false when using BOOST_ROOT and
set to true when useing other system directories like /usr/local
Also simplify using a set to remove duplicate lib files
Also remove the part where you search the Cellar in homebrew, this is
unnescessary now that symlinks can be followed it should find boost
when searching /usr/local so no need to search the Cellar
This does two things:
* allows the library files to be symlinks
* searches `lib` and `lib64` in `BOOST_ROOT` even if it finds lib
directories from the compiler
The first condition is needed for the homebrew on macOS because boost and boost
python are provided in seperate packages and are put together in /usr/local/lib
with symlinks to the library files. The both conditions are needed for high
performace computing environments where dependencies are often provided in
nonstandard directories with symlinks
A test case was added which looks for boost libraries in seperate directories
which have been symlinked to BOOST_ROOT/lib
GTestDependencySystem (and other similar dep classes) sets
self.is_found=True, but the version check could still fail. In the case
the dependency is not required `ExternalDependency._check_version()`
won't raise an exception and thus the dependency is accepted.
_check_version() sets self.is_found() in the case self.version is not
empty, we should do it too when self.version is empty.
Fixes: #9036.
Since intl links internally to iconv, try to pull in iconv too, but only
if we specifically ask for static libs.
If PREFER_SHARED, we have no way to tell which type we got, so do not
try handling that...
Although find_library returns the library as a list of args, has_header
returns a tuple(is_found, is_cached) which is non-empty and thus always
true. Which is confusing...
Check whether it actually got found.
Just like we automatically provide some reusable glue for self.static,
provide it here too. It seems plausibly like something people would
commonly want.
Both of these are artifacts of the time before Dependency Factories,
when a dependency that could be discovered multiple ways did ugly stuff
like finding a specific dependency, then replacing it's own attributes
with that dependency's attributes. We don't have cases of that left in
the tree, so let's get rid of this code too
* dependencies: Deterministic LLVM compile and link arg ordering
In LLVMDependencyConfigTool, the members compile_args and required_modules are
either converted to or stored as sets, which do not have a stable ordering. This
results in nondeterministic builds, particularly with required_modules causing
the order in which the LLVM libraries are linked in to the output binaries to
change across independent builds. As any guarantee about ordering for
compile_args is lost by being converted from a list to a set and back, and the
modules added to required_modules was even already sorted once, sort both when
converting them to lists.
* Use mesonlib.OrderedSet instead of sorting the sets.
Co-authored-by: Kaelyn Takata <kaelyn.alexi@protonmail.com>
This commit introduces a new type of `HoldableObject`: The
`SecondLevelHolder`. The primary purpose of this class is
to handle cases where two (or more) `HoldableObject`s are
stored at the same time (with one default object). The
best (and currently only) example here is the `BothLibraries`
class.
Checking how to aquire the *gettext family of symbols portably is
annoyingly complex, and may come from the libc, or standalone.
builtin dependency:
This detects if libintl is unneeded, because the *gettext family of
symbols is available in the libc.
system dependency:
This detects if libintl is installed as separate software, linkable via
-lintl; unfortunately, GNU gettext does not ship pkg-config files for
it.
Fixes#3929
mesonbuild.dependencies.__init__ exposes configtool, pkgconfig, cmake
and more in __init__.py, so there's no reason we should be tying
SystemDependency to the internal organization implementation of the
subpackage!
In the 2nd previous commit it took quite some effort to figure out that
the python module "does not exist" because of import errors while
refactoring something completely different.
For dependencies that on some systems are built into libc etc. and don't
need to be separately linked. This is distinct from "system"
dependencies which add linker args.
This allow mypy to catch cases where we accidently assign the dependency
name to the type_name, as it sees them as having different types (though
at runtime they're all strings).
Eli Schwartz
Dudemanguy
Jon Turney
Thomas Heijligen
Daniel Mensinger
Christian Clauss
Dylan Baker
Alan Coopersmith
Mat Cross
Paolo Bonzini
lilinzhe
Kevin Meagher
Xavier Claessens
kira78