This reverts commit 79c6075b56.
# Conflicts:
# docs/markdown/snippets/devenv.md
# mesonbuild/modules/python.py
# test cases/unit/91 devenv/test-devenv.py
PYTHONPATH cannot be reliably determined. The standard use case for
installing python modules with Meson is mixed pure sources (at least
`__init__.py`) and compiled extension_modules or configured files.
Unfortunately that doesn't actually work because python will not load
the same package hierarchy from two different directories, one a source
directory and one a (mandatory) out of tree build directory.
(It kind of can, but you need to do what this test case accidentally
stumbled upon, which is namespace packages. Namespace packages are a
very specific use case and you are NOT SUPPOSED to use them outside that
use case, so people are not going to use them just to circumvent Meson
devenv stuff as that would have negative install-time effects.)
Adding PYTHONPATH anyway will just lead to documentation commitments
which we cannot actually uphold, and confusing issues at time of use
because some imports *will* work... and some will *not*. The end result
will be a half-created tree of modules which just doesn't work together
at all, but because it partially works, users attempting to debug it
will spend time wondering why parts of it do import.
For any case where the automatic devenv would work correctly, it will
also work correctly to use `meson.add_devenv()` a single time, which is
very easy to manually get correct and doesn't provide any significant
value to automate.
In the long run, an uninstalled python package environment will require
"editable installs" support.
Dependencies in the "if_true" keyword argument do not prevent the
sources from being used; in other words, they work just like dependencies
with "disabler: false".
However, this was broken in commit ab0ffc6a2 ("modules/sourceset: Fix
remaining typing issues", 2022-02-23) which changed logic instead of
just fixing typing issues. This was likely an attempt to avoid using
"dependencies.Dependency" after the "dependencies" field was declared,
but it also broke QEMU.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The dependencies field clashes with the dependencies module, so that
mypy interprets "dependencies.Dependency" as a "Dependency" attribute
of the "dependencies" field.
Rename the field to something else, so that it does not clash.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
In commit 68e684d51f the _get_link_args
function was modified from returning a list[str] of arguments, to a
tuple of both that and a modified copy of the entire target's
current/enhanced dependencies (why not just the new ones? I don't know).
However, the existing use of the function was not adapted to this
change, and tried to turn this entire tuple into a node of the command
line. Tuples cannot flatten to lists, and mesonlib.File or
HoldableObjects don't make good command line arguments.
As a result we errored out with:
ERROR: Argument (['-L/path/to/builddir/', '--extra-library=foo'], [<SharedLibrary 25a6634@@foo@sha: foo>, <SharedLibrary 25a6634@@foo@sha: foo>, <SharedLibrary 25a6634@@foo@sha: foo>]) in "command" is invalid
Split out the flags and the dependencies and update the former while
replacing the latter.
In commit c88bfdbefc we added support for
an exe_wrapper to gtkdoc, which checked twice whether the environment
says it is needed, and didn't check at all whether one was provided.
The result:
File "/usr/lib/python3/dist-packages/mesonbuild/modules/gnome.py", line 1354, in gtkdoc
t_args.append('--run=' + ' '.join(state.environment.get_exe_wrapper().get_command()))
AttributeError: 'NoneType' object has no attribute 'get_command'
Instead, check whether we have a valid exe_wrapper (if we don't need
one, then even when one is defined in the cross file, we get an
EmptyExternalProgram) and if we do, use it.
If we don't have one, but need one, then we revert back to the behavior
before commit c88bfdbefc, which probably
means "executing the doc target causes the command to error out with
"Exec format error".
There are two cases where we can assume we found the python dependency
with its requisite libraries using sysconfig:
- we found the library with find_library and are prepared to link to it
- the library is not actually part of the dependency, so its presence or
absence is irrelevant
In the latter case, we should consider it found if link_libpython is
False. Originally we did this, but the logic was inverted in commit
5b422fce87 in an unrelated change and
without explanation, likely by accident.
Normally this doesn't much matter, since a python invariably comes with
a predictably located libpython and the first condition evaluates true.
But that is not true for pypy, and in fact that is the reason the
link_libpython check was originally added in commit
1bd14b52b2.
Restore that original logic.
Fixes#8570
Using future annotations, type annotations become strings at runtime and
don't impact performance. This is not possible to do with T.cast though,
because it is a function argument instead of an annotation.
Quote the type argument everywhere in order to have the same effect as
future annotations. This also allows linters to better detect in some
cases that a given import is typing-only.
A bunch of files have several T.TYPE_CHECKING blocks that each do some
things which could just as well be done once, with a single `if`
statement. Make them do so.
Disabling targets because the tools used to build them aren't available
is a pretty suspicious thing to do. Users who want this are probably, in
general, advised to check themselves whether it is possible to build
those targets with find_program(..., required: false)
The i18n.gettext() invocation is a bit unusual because the product of
running it is non-critical files, specifically, translation catalogs. If
users don't have the tools needed to build them, they may not be able to
use them either, because perhaps they have NLS disabled on their
platform or it's difficult to put it in the bootstrap path.
So, for this reason, it was made non-fatal and the message catalogs are
just not created, and the resulting build is still perfectly usable
*unless* you want to use it in another language, at which point it
"works" but the text is all inscrutable to the end user, and that's a
feature of the target platform.
That's an acceptable tradeoff for translation catalogs.
It is NOT an acceptable tradeoff for merge_file, which produces desktop
files or MIME database catalogs or other files which have crucial roles
to perform, without which the software in question simply doesn't work
at all. In such cases, this just fails to install crucial files, users
report bugs to the project in question, and the project adds
`find_program('xgettext')` to guarantee the hard error due to lack of
confidence in Meson.
Fixes#6165Fixes#8436
Use this instead of shutil.which to detect whether they will be
available, and pass the ExternalProgram object to CustomTarget
invocations, or else make use of the new functionality to specify the
correct program path in wrapper scripts.
Drop duplicate reporting for itstool missing. Since we use find_program
in required mode, its absence is already fatal, and already has a really
good error description.
Due to misuse of argparse in commit 82492f5d76
it was impossible to use both --datadirs and extra args passed directly
to msgfmt at the same time.
I'm not sure anyone actually knows how argparse works, so misusing it is
easy. What is definitely known is that argparse is NOT a POSIX compliant
parser and doesn't behave the way you'd expect a standards based parser
to handle options. Instead it caters to the easy use case, and hopes and
prays you don't need to do anything too complicated "with the wrong kind
of complicated".
Apparently, this particular type of complicated is when you have mixed
option_arguments and operands while simultaneously passing some operands
as nargs after a --.
It totally breaks, and interprets --datadirs, which is supposed to be an
option_argument, as an operand, eats it up as a msgfmt wrapped argument,
and breaks.
But if you don't pass additional arguments with -- then it interprets
--datadirs after operands as an option_argument. This is what we were
doing.
Instead pass option_arguments before all operands (including the ones
specified via `-- ...`). Add test case to pass meaningless datadirs (we
don't actually care if $GETTEXTDATADIRS is set to something that doesn't
contain gettext data).
- Change `scope` kwarg to `public` boolean default to false.
- Change `side` kwarg to `client` and `server` booleans.
- Document returned values
- Aggregate in a single unit test because have lots of small tests
increases CI time.
Fixes: #10040.
After implementing a much more extensive Java native module than what
currently exists in the tests, I found shortcomings.
1. You need to be able to pass multiple Java files.
2. Meson needs more information to better track the generated native
headers.
3. Meson wasn't tracking the header files generated from inner classes.
This new function should fix all the issues the old function had with
room to grow should more functionality need to be added. What I
implemented here in this new function is essentially what I have done in
the Heterogeneous-Memory Storage Engine's Java bindings.
The point of a .use() function is because we don't always have the
information we need to use a feature check, so we allow creating the
feature and then storing it for later use. When implementing location
checks, although it is optional, actually using it violated that design.
Move the location out of the init method for FeatureCheck itself. It
remains compatible with all cases of .single_use(), but fix the rest up.
The default behavior of installing relative to prefix may be unexpected,
and is definitely wrong in many cases.
Give users control in order to specify that yes, they actually want to
install to a venv.
This is particularly useful for projects that use meson as a build
system for a python module, where *all* files shall be installed into
the python site-packages.
Automatically generate additional variables and write them into the
generated pkg-config file.
This means projects no longer need to manually define the ones they
use, which is annoying for dataonly usages (it used to forbid setting
the base library-relevant "reserved" ones, and now allows it only for
dataonly. But it's bloat to manualy list them anyway).
It also fixes a regression in commit
248e6cf473 which caused libdir to not be
set, and to be unsettable, if the pkg-config file has no libraries but
uses the ${libdir} expansion in a custom variable. This could be
considered likely a case for dataonly, but it's not guaranteed.
CMake's write_basic_package_version_file has supported since version 3.14
an ARCH_INDEPENDENT option that makes it skip its architecture check in
the Version file.
With this patch Meson now supports it as well, and the change is also
compatible with older CMake versions, as they will simply ignore the
option.
This also slightly changes the contents of the generated Version file
when arch_independent is not set: previously, the if() needed to skip
the arch check was always filled with an empty string, while CMake puts
"FALSE" (or "TRUE") in it. Now, that if() will always be filled with
either "False" or "True", better matching CMake's behaviour.
Eli Schwartz
Paolo Bonzini
Ralf Gommers
Tristan Partin
Dylan Baker
Xavier Claessens
Remi Thebault
Mark Bolhuis
Andrea Pappacoda