A later commit will add a second invocation of ninja, no point in having the
detection code twice.
This changes the exit code in case ninja isn't found from 125 to 127, which
seems more appropriate given the justification for returning 125 (to make git
bisect run skip that commit). If we can't find ninja we'll not succeed in
other commits either.
A subsequent commit will do a bit more during test data loading, making a
dedicated function seem advisable.
The diff looks a bit odd, using git show --diff-algorithm=patience
will make it clearer.
Currently a cosmetic bug is present: once a build dir was regenerated,
meson would start showing:
User defined options
backend: ninja
This is not true as user have not defined the option, it is default.
Fix this by omitting the `--backend ninja` parameter from "regenerate"
In my tests this does not affect the situation when one specifies
`--backend ninja` explicitly, it still shows the backend as user-defined
after reconfiguration.
Fixes: https://github.com/mesonbuild/meson/issues/10632
Meson internally knows about many languages and tools, and *FLAGS
variables, and which languages to use them for. Instead of duplicating
this logic, import it from mesonbuild.*
This logic was originally standalone, but now that it is merged into the
Meson tree we can have a single source of truth.
This can be triggered if someone tries to call a non-ID. The example
reproducer was:
```
if (var = dependency(...)).found()
```
This produced a traceback ending in
```
raise InvalidArguments(f'Variable "{object_name}" is not callable.')
UnboundLocalError: local variable 'object_name' referenced before assignment
```
After this commit, the error is reported as:
```
ERROR: AssignmentNode is not callable.
```
We use `__future__.annotations` to good effect everywhere we can, and
one of the effects of this is that annotations are automatically
stringized and don't need to be evaluated, using less memory and
computation. But this only affects actual annotations -- a cast is just
a function with an argument, so the compiler has no idea that it's an
annotation to be stringized.
Do this manually.
This is supposed to expose all primitives together, but to do that we
need to actually "use" each variable in `__all__`, which we... didn't.
Sorry about that.
Strictly speaking code restructuring isn't needed, but making this PEP8
compliant results in indentation of the code that reduces the
readability. By moving the offending code on the outside of the method
call, the readability is maintained.
What happens is this:
- liba is a convenience static library
- libb is an installed static library
- libb links in liba with --link-whole
- libc links to libb
- we generate a link line with libb *and* liba, even though libb is a
strict superset of liba
This is a bug that has existed since the we stopped using link-whole to
combine convenience libraries, and to instead propagate their
dependencies up. For most linkers this is harmless, if inefficient.
However, for apple's ld64 with the addition calling `ranlib -c`, this
ends up causing multiple copies of symbols to clash (I think that other
linkers recognize that these symbols are the same and combine them), and
linking to fail.
The fix is to stop adding libraries to a target's `link_whole_targets`
when we take its objects instead. This is an all around win since it
fixes this bug, shortens linker command lines, and avoids opening
archives that no new symbols will be found in anyway.
There are two distinct cases here that need to be considered. The first
issue is https://github.com/mesonbuild/meson/issues/10723 and
https://github.com/mesonbuild/meson/issues/10724, which means that Meson
can't actually generate link-whole arguments with rust targets. The
second is that rlibs are never valid candidates for link-whole anyway.
The promotion happens to work because of another bug in the promotion
path (which is fixed in the next commit).
This is generally a bad idea, e.g. it causes OSError on freebsd.
It also gets ignored by solaris and thus causes unittest failures.
The proper solution is to simply reject any attempt to set this, and log a
warning.
The install_emptydir function does apply the mode as well, and since it
is a directory it actually does something. This is the only place where
we don't reset the mode.
Although install_subdir also installs directories, and in theory it
could set the mode as well, that would be a new feature. Also it doesn't
provide much granularity and has mixed semantics with files. Better to
let people use install_emptydir + install_subdir.
Fixes#5902
Generally plumb through the values of get_option() passed to
install_dir, and use this to establish the install plan name. Fixes
several odd cases, such as:
- {datadir} being prepended to "share" or "include"
- dissociating custom install directories and writing them out as
{prefix}/share/foo or {prefix}/lib/python3.10/site-packages
This is the second half of #9478Fixes#10601
The name can be None if a library is not passed as a positional
argument, and the name keyword argument is not provided. We shouldn't
allow that to happen.
There are still a lot of errors here due to a mixture of really bad
design (adding extra attributes to objects), and legitimate type errors
(passing a str where a List[str] is expected). I suspect a lot of these
cases aren't hit for some reason.
`configure_file` is both an extremely complicated implementation, and
a strange place for copying. It's a bit of a historical artifact, since
the fs module didn't yet exist. It makes more sense to move this to the
fs module and deprecate this `configure_file` version.
This new version works at build time rather than configure time, which
has the disadvantage it can't be passed to `run_command`, but with the
advantage that changes to the input don't require a full reconfigure.
install_mode can include the setuid bit, which has the special property
(mentioned in the set_mode logic for minstall itself) of needing to come
last, because it "will get wiped by chmod" (or at least chown).
In fact, it's not just chown that wipes setuid, but other changes as
well, such as the file contents. This is not an issue for install_data /
custom_target, but for compiled outputs, we run depfixer to handle
rpaths. This may or may not cause edits to the binary, depending on
whether we have a build rpath to wipe, or an install rpath to add. (We
also may run `strip`, but that external program already has its own mode
restoration logic.)
Fix this by switching the order of operations around, so that setting
the permissions happens last.
Fixes https://github.com/void-linux/void-packages/issues/38682
Do to a bug in which a method, rather than an attribute, was checked
we would never detect that a target had no sources, even though we meant
to. As a result we would never actually error out when a target has no
sources. When fixing the check, I've changed the error to a warning
so that existing projects will continue to build without errors, only
warnings.
This was never meant to work, it's an implementation detail of using
`importlib.import_module` and that our modules used to be named
`unstable_` that this ever worked.
Thanks to `ModuleInfo`, all modules are just named `foo.py` instead of
`unstable_foo.py`, which simplifies the import method a bit. This also
allows for accurate FeatureNew/FeatureDeprecated use, as we know when
the module was added and if/when it was stabilized.
We're going to do more with this in the next commit, but this just adds
the information for now. This allows the next commit have 100% mv
changes for some of the modules, which makes review easier
Instead of using FeatureNew/FeatureDeprecated in the module.
The goal here is to be able to handle information about modules in a
single place, instead of having to handle it separately. Each module
simply defines some metadata, and then the interpreter handles the rest.
Andres Freund
Konstantin Kharlamov
Jehan
Eli Schwartz
aleksander
Samay Sharma
Alf Henrik Sauge
Fabian Orccon
Dylan Baker
Xavier Claessens