Reorder meson targets to handle those all at the end, and exit early if
HTML documentation is disabled. This makes it possible to build just the
manpage, without hotdoc installed.
CPython adds a hard to avoid hardcoded link flag to look for a debug
libpython. This results in a very puzzling error, so emit a warning
with a clear message. Note that pybind11 has a workaround for this,
which undefines `_DEBUG`. So users who use only pybind11 can use
non-release buildtypes, but they won't get debug symbols on Windows
unless they have a debug build.
This matches the tests for Python extensions.
Also include some other cleanups to these `meson.build` files:
Adding `python_dep` is no longer needed, this is automatic now.
Use a single line for `import('python').find_installation()`,
because the result of `import('python')` by itself is not used
for anything.
If the dependency permits it, we can just do a PATH search instead of
mandating that it be listed in the cross file. This is useful for the
limited case where a specific dependency is known to be compatible with
any machine choice.
Mark the pybind11 dependency as supporting this. It's a valid choice
because pybind11 is a header-only C++ library.
In #11761 it turned out that we failed to correctly handle all
compiler.sizeof API changes in an old commit, breaking use of the
module. And mypy could have caught this for us, except that the module
is neither typed nor checked in CI. Partially solve this by adding lots
of type annotations, greatly reducing the number of mypy errors in this
file from 35 down to 12.
This doesn't accept a dict, only an actual ConfigurationData object. Due
to the way we poke at it, a dict can sort of work anyway, but might not
if the internal layout isn't exactly correct. This is evidenced by the
way we make the dict values be hard-to-read tuples containing emptiness,
because that's how ConfigurationData objects handle descriptions.
Simplify and make the seed dictionary readable, then actually convert it
into a real ConfigurationData. Bonus: this now passes type checking.
It's shorter and more descriptive. Although we always enforce the same
rules either way, a unified decorator is one less line of code for each
location, and also tells you how many "too few" arguments you *did*
pass.
The stdlib unittest module has a magic flag (undocumented) which
indicates that a module is part of a unittest framework.
> Truncates usercode tb at the first unittest frame.
>
> If the first frame of the traceback is in user code,
> the prefix up to the first unittest frame is returned.
> If the first frame is already in the unittest module,
> the traceback is not modified.
This avoids some ugliness, e.g. the following test error logs:
```
> self.assertPathListEqual(intro[0]['install_filename'], ['/usr/lib/libstat.aaa'])
unittests/allplatformstests.py:432:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
unittests/baseplatformtests.py:393: in assertPathListEqual
self.assertPathEqual(i[0], i[1])
unittests/baseplatformtests.py:384: in assertPathEqual
self.assertEqual(PurePath(path1), PurePath(path2))
E AssertionError: PurePosixPath('/usr/lib/libstat.a') != PurePosixPath('/usr/lib/libstat.aaa')
```
Since assertPathListEqual is our own assertion helper, we don't need to
give trace information about its internals. This change causes the error
log to become:
```
> self.assertPathListEqual(intro[0]['install_filename'], ['/usr/lib/libstat.aaa'])
E AssertionError: PurePosixPath('/usr/lib/libstat.a') != PurePosixPath('/usr/lib/libstat.aaa')
unittests/allplatformstests.py:432: AssertionError
```
which is a lot more readable.
msys2 is broken only on clang, due to -Werror issues in the python
headers as patched by msys2.
MSVC is simply weird... due to the use of an unversioned platlib/purelib
directory, the python2 and python3 components overlap.
For all source `*.py` files installed via either py.install_sources() or
an `install_dir: py.get_install_dir()`, produce `*.pyc` files at install
time. Controllable via a module option.
We may want to do things like update install scripts as well, which have
to happen before generating the backend. Instead of adding one module
method per thing to do, use a single function that allows for modifying
the Build object directly.
This is useful for internal scripts that want to know about something
other than MESON_INSTALL_PREFIX and MESON_INSTALL_DESTDIR_PREFIX, which
is very specific to the prefix.
In commit 808d5934dd, compiler.sizeof was
refactored to introduce caching, but cmake subprojects did not adapt to
that API change and ended up embedding the python repr of a tuple as a
cmake variable.
In this case, we have the secret available, and the workflow ran even
though it wasn't on branch "master" because of the pull request trigger.
Since the change hasn't landed on master, though, we do not want to
update the website. So check for pushes to master, specifically.
To take good decisions we'll need to know if we are a Rust library which
is only know after processing source files and compilers.
Note that is it not the final list of compilers, some can be added in
process_compilers_late(), but those are compilers for which we don't
have source files any way.
Case 1:
- Prog links to static lib A
- A link_whole to static lib B
- B link to static lib C
- Prog dependencies should be A and C but not B which is already
included in A.
Case 2:
- Same as case 1, but with A being installed.
- To be useful, A must also include all objects from C that is not
installed.
- Prog only need to link on A.
patch on Windows is provided by MSYS, which only understands POSIX
paths, with `/`. Using Windows paths with `\` results in a "file not
found" error.
We got a little lucky here because the path is relative, so the drive
letter difference doesn't affect us.
MIPS64 can run MIPS32 code natively, so there is a chance that a mixture
of MIPS64 kernel and MIPS32 userland exists. Before this Meson just
treats such mixture as mips64, because uname -m returns mips64.
So in this case we have to check compiler builtin defines for actual
architecture and CPU in use.
- Also fixes mips64 related detection tests in internaltests:
Normalize mips64 as mips first, then if __mips64 is defined, return
mips64 for mips64* machines.
This is a bit confiusing because normally one would detect if a flag
of 32-bit target is defined while running on a 64-bit machine. For
mips64 it is almost just the other way around - we need to detect if
__mips64 is set to make sure it is a mips64 environment.
Co-Authored-By: Jue Wang <maliya355@outlook.com>
Eli Schwartz
Nomura
Volker Weißmann
Jussi Pakkanen
Joel Rosdahl
Xavier Claessens
Radek Bartoň
Ralf Gommers
Renan Lavarec
Benoit Pierre
MaxHearnden
Charles Brunet
Nirbheek Chauhan
Cyan
Peter Hull