Unfortunately, checking for strings without context is exceedingly prone
to false positives, while missing anything that indirectly opens a file.
Python 3.10 has a feature to warn about this though -- and it uses a
runtime check which runs at the same time that the code fails to open
files in the broken Windows locale. Set this up automatically when
running the testsuite.
Sadly, Python's builtin feature to change the warning level, e.g. by
setting EncodingWarning to error at startup, is utterly broken if you
want to limit it to only certain modules. This is tracked in order to be
more efficiently ignored at https://bugs.python.org/issue34624 and
https://github.com/python/cpython/pull/9358
It is also very trigger happy and passing stuff around via environment
variable either messes with the testsuite, or with thirdparty programs
which are implemented in python *such as lots of gnome*, or perhaps
both.
Instead, add runtime code to meson itself, to add a hidden "feature".
In the application source code, running the 'warnings' module, you can
actually get the expected behavior that $PYTHONWARNINGS doesn't have. So
check for a magic testsuite variable every time meson starts up, and if
it does, then go ahead and initialize a warnings filter that makes
EncodingWarning fatal, but *only* when triggered via Meson and not
arbitrary subprocess scripts.
We didn't consider that it has arguments following it, so the resulting
compiler command line ended up with stuff like:
-L=-rpath-link -L=-L=/path/to/directory -L=more-args
and the directory for rpath-link got eaten up as a regular -L path to
the compiler rather than being passed as -Xlinker to the linker.
Then the -rpath-link would consume the next -Xlinker argument, end up
with the wrong rpath-link (may or may not cause link errors) and then
disappear arguments we need.
As an example failure mode, if the next argument is -soname this treats
the soname text as an input file, which probably does not exist if it
was generated in a subdirectory, and also because it can never be
successfully built in the first place -- though if it did, it would link
to itself which is very wrong.
If --no-rebuild is used, the test program might not exist, spawning a
FileNotFoundError inside a long traceback rooted in subprocess.Popen
trying to run that test program. Current versions of Meson even say it's
an "unhandled python exception".
But we can do one better and actually tell the user what is wrong, why,
and what to do to fix it. And we can do so before getting waist deep in
partially running tests.
Fixes#10006
In a bunch of cases we create a series of sample libraries named "a",
"b", "c" etc.
This breaks on musl. Originally reported with muon via commit
ca5c371714
and also breaks the testsuite when packaging meson for alpine linux.
libc.so is an existing library which is linked in by default for all the
obvious reasons. You can get away with this on glibc, because that
includes a soversion of "6", but it loads the wrong library on musl.
Follow-up on commit 4274e0f42a. We want to
allow tests to be skipped freely in third-party environments, so this
should check the jobname, not whether $CI exists.
We will anyways raise an error when trying to run the project tests, if
$CI is set but no jobname is set.
This header is required anyway. And the compile test for linking to libc
with the gettext symbol, can succeed when we try to use the literal
symbol name without includes, but fail later during project build,
because actually including libintl.h might redefine the function to match
a forked symbol. This happens when GNU libintl is installed as a
standalone library on systems that have a less fully-featured gettext
implementation.
So, by including the header in has_function, we can ensure that we test
against the default resolved header. In the event that the symbol which
is #define'd by the header is 'libintl_gettext', linking will fail against
libc even when a builtin gettext does exist, and we will fall back to the
intl dependency that provides -lintl (and which is needed to properly
use the default header).
Of course, even that probably won't work.
has_function(prefix: '...') is useless to check the difference between
builtins and external library functions. It has code to detect
"builtins" that misfires in some cases (previously seen with iconv_open).
Instead compile an open-coded test file that this intl dependency
implementation fully controls, that doesn't get up to imaginative edge
cases like trying to find `__builtin_gettext`.
It's the only way to be sure.
Fixes compiling against the intl dependency on e.g. Alpine Linux when the
libintl package is installed.
This is treated by the test harness as though unset, i.e. we do normal
skipping and don't assume we are running in Meson's own project CI.
However, it has one distinction which is that it isn't an error to set
$CI without setting $MESON_CI_JOBNAME, if it is in fact set but to the
ignored value.
This lets automated workflows such as Linux distro testing, particularly
alpine linux, set $CI or have it set for them by default, without
messing things up.
Also it has the advantage of $CI actually enabling useful benefits! We
will still assume that this thirdparty environment wants to force
verbose logging (printing testlogs, running ninja/samu with -v) and
colorize the console.
find_program() can check that for us already, there's no need to
require GNU bash on systems that default to other POSIX shells.
Even though this is the *linuxlike* tests and a POSIX shell is
guaranteed, there is actually no need to require a shell at all. It's
*easier* to use the meson builtin functionality here.
Instead of blindly assuming when $CI is set that `ninja` is always
correct and always new enough, check for it the same way we do when $CI
is not set.
Instead of special casing when $CI is set and skipping ninja detection
in subprocess tests (by assuming it is always `ninja`), skip detection
in subprocess tests all the time, by passing this information around
across processes after the first time it is looked up.
This means that local (non-CI) tests are faster too!
Fixes running unittests in alpine linux using samu. Although ninja is a
symlink to samu, the exact output string on no-op builds is different,
which we already handle, but what we don't handle is the fact that samu
prints a third case when you invoke it as `ninja`. If $CI is exported,
then the unittests ignored $NINJA entirely.
Fixes running unittests when $CI is set, `samu` exists and `ninja` does
not exist. This is not currently the case anywhere, but may be in the
future; why not fix it now?
Android requires shared modules that use symbols from other shared
modules to be linked before they can be dlopen()ed in the correct
order. Not doing so leads to a missing symbol error:
https://github.com/android/ndk/issues/201
We need to always allow linking for this. Also add a soname, although
it's not confirmed that it's needed, and it doesn't really hurt if it
isn't needed.
On Windows using MSYS2 MinGW installing the package `mingw-w64-x86_64-wxmsw3.1` provides `wx-config-3.1`. I have tried building my software by making this exact change and it build correctly.
Do not quit from using pytest at all, when pytest is present, simply
because xdist isn't available. Even without xdist, pytest is still
useful.
There doesn't seem to be any particular reason to require xdist. It just
happens to have been implemented that way, back in commit
4200afc74d when we originally added a
check to avoid pytest erroring out with unknown options when xdist
options are passed and xdist is not installed.
The $CI environment variable may be generally set by Github or Gitlab
actions, and is not a reliable indicator of whether we are running "CI".
It could also, for an absolutely random example that didn't *just
happen*, be Alpine Linux's attempt to enable the Meson testsuite in
their packaging, which... uses Gitlab CI.
In this case, we do want to perform normal skipping on not-found
requirements. Instead of checking for $CI, check for $MESON_CI_JOBNAME
as we already use that in all of our own CI jobs for various reasons.
This makes it easier for linux distros to package Meson without
accumulating hacks like "run the testsuite using `env -u CI`".
Fixes regression in commit 05b5a1e56f.
This added usage of another module in the python module's introspection
of `meson.exe runpython`, but the MSI packaging didn't adapt, causing it
to fail to be detected due to ImportError.
Fixes#9975
In commit b30dddd4e5, various refactorings
were done, during which a kwarg got accidentally dropped from the
function that determined part of the log message. As a result, a ':'
suddenly appeared in the log message where none should be.
Example expected output:
Checking if "-Werror=shadow with local shadowing" compiles: YES
What actually happened:
Checking if "-Werror=shadow with local shadowing" : compiles: YES
Fixes#9974
RuntimeError is way too generic when we have an explicit class for
"Meson reports to the user, something went wrong".
Moreover we now tell people that generic exceptions are "Meson bugs and
should be reported", so our failure to do the technically correct thing
and report the right kind of exception means we get haunted by demons of
confusion. Specifically, people complain to us that Meson told them
"there is a bug in Meson" when their install fails due to meson.build or
build environment issues.
This will really help with pipx run, as `pipx run meson[ninja]` will now work on any system with pipx 0.17 or newer, no dependencies required. This now includes GitHub Actions, which just updated to pipx 1.0.0. Pipx run lets you use recent (always <1 week old) versions of anything on PyPI.
This has never worked for built/found programs, only for script files.
In commit 2fabd4c7dc scripts learned an
attribute stating which subproject they came from. In commit
3990754bf5 dist scripts learned to run
even from a subproject, and relied on that attribute to know when, in
fact, they came from a subproject.
Unfortunately the original attribute was only set in one half of an
if/else, and the other half returned early with only part of the work
done.
Fixes#9964
In commit 0deab2ee9e we added the ability
to pass a declare_dependency() to any compiler method that accepts
"dependencies", but we never marked the version it is available since.
Fixes#9957
For static library crates that depend on other internal static library
crates, all link_with targets get promoted to link_whole targets. Due to
a bug, only link_with targets are considered when generating a Rust
target for the ninja backend. This made it impossible to link a Rust
static library with another internal Rust static library.
This change fixes that issue by chaining link_whole_targets with
link_targets, just like many other languages within the ninja backend.
Basically the last thing we did during target processing was to generate
shlib symlinks for e.g. libfoo.so -> libfoo.so.1.
In some cases we would dispatch to another function and return early,
though, which meant we never got far enough to generate the symlinks.
This then led to breakage when people tried to compile against libfoo.so
This surely breaks -uninstalled.pc usage, and also caused problems in
https://github.com/rust-lang/rust/pull/90260
Backported from 0185f2ed61, which contains
a fix that doesn't need to be backported, but we can backport the test
to ensure that it's still working in the 0.61 branch.
unittests/rewritetests.py:19: DeprecationWarning: The distutils package is deprecated and slated for removal in Python 3.12. Use setuptools or check PEP 632 for potential alternatives
from distutils.dir_util import copy_tree
For example:
```
meson builddir \
--native-file vs2019-paths.txt \
--native-file vs2019-win-x64.txt \
--cross-file vs2019-paths.txt \
--cross-file vs2019-win-arm64.txt
```
This was causing the error:
> ERROR: Multiple producers for Ninja target "/path/to/vs2019-paths.txt". Please rename your targets.
Fix it by using a set() when generating the list of regen files, and
add a test for it too.
The utility function that processes this for both 'variables' and
'uninstalled_variables' accepts a kwarg for the name of the argument,
but then hardcodes 'variables' in the warning message. This is
misleading.
It's not a MesonBug which needs to be reported, and the existing error
already adequately points out the problematic file.
It is impossible to get a PermissionError for files created by meson
itself, once the build directory has been created, anyway.
This was allows up to 0.61.0 (including with the initial type
annotations), but was accidentally broken by fixes for other bugs in
0.61.1.
Fixes: #9883
In commit 06481666f4 this warning got
moved from build.py to the interpreter. Unfortunately it got added to
the wrong function... it is supposed to be part of custom_target and
even mentions this as the feature_name.
Since then, build_always became a KwargInfo and has the deprecated-since
attribute baked into it. But it didn't have the additional message which
it really should have.
Add that message at the same time we remove it from vcs_tag.
In commit 928078982c a good error message
about the directory not being a valid build directory, was replaced by a
worse message.
In commit abaa980436 the error message was
replaced by a traceback when trying to load the coredata before checking
if it was a build directory.
Revert back to using the helper function with the good error message.
Reorganize code so that we check basic things first, and do less work
before detecting errors.
Fixes#9584
Now cygwin seems to have completed a migration of the default python to
3.9, so that is where the devel package is at.
Back out the changes from commit 3304a38496
and update the pip/wheel packages as appropriate.
The tests and the unittests both unconditionally call setup_vsenv()
because all tests are run using the backend commands directly: ninja,
msbuild, etc.
There's no way to undo this vs env setup, so the only way to test that
--vsenv works is by:
1. Removing all paths in PATH that provide ninja
2. Changing setup_vsenv(force=True) to forcibly set-up a new vsenv
when MESON_FORCE_VSENV_FOR_UNITTEST is set
3. Mock-patching build_command, test_command, install_command to use
`meson` instead of `ninja`
4. Asserting that 'Activating VS' is in the output for all commands
5. Ensure that compilation works because ninja is picked up from the
vs env.
I manually checked that this test actually does fail when the previous
commit is reverted.
Otherwise we might not find a ninja that was picked up from the Visual
Studio installation.
```
$ meson setup _build
...
Activating VS 15.9.40
...
Found ninja-1.8.2 at "C:\Program Files (x86)\Microsoft Visual Studio\2017\Community\Common7\IDE\CommonExtensions\Microsoft\CMake\Ninja\ninja.EXE"
$ meson compile -C _build
Activating VS 15.9.40
...
$ meson install -C _build
Can't find ninja, can't rebuild test.
```
Fixes https://github.com/mesonbuild/meson/issues/9774
It used to support:
- a single string
- an array of anything
And as long as CustomTarget supported it too, everything worked fine.
So, a `files('foo')` worked but a `files('foo')[0]` did not, which is
silly... and it's not exactly terrible to use files() here, the input is
literally a list of source files.
Fixes building gnome-terminal
Fixes#9827
Test updated by Nirbheek Chauhan <nirbheek@centricular.com>
check: true is needed for an actual regression test, but that requires
a pedantically correct gtk-doc configuration, which I attempted to do
and failed. So let's just put check: false so we get *some* coverage:
just that typed_kwargs accepts the argument.
Related to https://github.com/mesonbuild/meson/pull/9807
The original attempted fix only allowed configuration to succeed, but
not building. It was modeled based on the gdbus-codegen documentation,
which states:
--annotate WHAT KEY VALUE WHAT KEY VALUE WHAT KEY VALUE
Add annotation (may be used several times)
which clearly indicates that gdbus-codegen accepts an --annotate flag
that is followed by a multiple of 3 values, despite this not actually
working.
The manpage actually contradicts the --help text:
--annotate ELEMENT KEY VALUE
Used to inject D-Bus annotations into the given XML files. []
... and gives examples that use multiple --annotate flags each with 3
arguments.
To better understand what meson is supposed to do here, we should look
at ef52e60936, which ported to
typed_kwargs. There is actually a big chunk of code to handle
annotations that got completely dropped, leading with a comment (that
did not get dropped): "they are a list of lists of strings..."
Reimplement this logic inside a validator/converter for the annotations
kwarg container:
- do not listify, we don't accept `annotations: ''` and listify is
supposed to be for when either x or list[x] is valid
- go back to checking for a list of exactly 3 values
- allow a list of the aforementioned, in the traditionally expected
form:
[
['foo1', 'foo2', 'foo3'],
['bar1', 'bar2', 'bar3'],
]
- pass one --annotate flag per 3-value-list
And add some better error reporting for the cause of errors when
processing lists of lists.
Visual Studio Express does not come with the
'Microsoft.VisualStudio.Component.VC.Tools.x86.x64' workload.
This adds a check for the 'Microsoft.VisualStudio.Workload.WDExpress'
workload. Non-express versions take precedence over express versions
when activating.
Eli Schwartz
Nirbheek Chauhan
apoorv569
Alyssa Ross
Henry Schreiner
Tobias Pankrath
Ole André Vadla Ravnås
Zach Reizner
Dylan Baker
Bilal Elmoussaoui
Gatgat
DFOVIT