| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Otherwise bindgen may generate different behavior than the compiled C
code actually has.
|
|
Rust has a `debug_assert!()` macro, which is designed to be toggled on
the command line. It is on by default in debug builds, and off by
default in release builds, in cargo. This matches what meson's b_ndebug
option does in `if-release` mode.
|
|
|
|
C like compilers only off `-DNDEBUG` to disable asserts. This is not a
universal paradigm however. Rust, for example has an argument that takes
a boolean. To better represent this, we allow passing a `disable`
boolean. `disable` was chosen rather than `enable` because it allowed
all existing logic to be left in place
|
|
The pypi package was suddenly removed. Not the most
confidence-inspiring deprecation/migration:
https://about.codecov.io/blog/message-regarding-the-pypi-package/
|
|
This will help with the writing of tools to generate
VisualStudio project and solution files, and possibly
for other IDEs as well.
- Used compilers a about `host`, `build` and `target` machines
arere listed in `intro-compilers.json`
- Informations lister in `intro-machines.json`
- `intro-dependencies.json` now includes internal dependencies,
and relations between dependencies.
- `intro-targets.json` now includes dependencies, `vs_module_defs`,
`win_subsystem`, and linker parameters.
|
|
|
|
|
|
The library names are directly mapped to filenames by meson while the
crate name gets spaces/dashes replaced by underscores. This works fine
to a certain degree except that rustc expects a certain filename scheme
for rlibs that matches the crate name.
When using such a library as a dependency of a dependency compilation
will fail with a confusing error message.
See https://github.com/rust-lang/rust/issues/110460
|
|
They're only used at build-time and whether they exist or not should not
have any effect on the final build artifact.
|
|
This reverts commit 0022e1863d8d9769378ebfb56695bdbf55f118b9.
The bug is long since both fixed and *released*.
|
|
cache/restore and cache/save now exist, and close the issue linked in
the workflow comment. The new save action runs when invoked, rather than
as a post action.
|
|
If the optional first "mainlib" argument is there, then we infer several
values. Otherwise, some of those values fall back to a generic default,
and two of them -- name and description -- fall back to being mandatory.
In commit e84f293f672a372d2434d0ce4fa39d3f902b6ce8, we removed
validation for description as part of refactoring that never actually
validated anything.
|
|
Like every single other command in the entire workflow, and for the same
reasons. ;)
|
|
|
|
|
|
Checking the executable basename sort of works, at least for Windows,
since Windows always happens to use exactly this approach. However, the
official pyinstaller documentation suggests a very different approach:
https://pyinstaller.org/en/stable/runtime-information.html
This approach is more robust since it works on any OS, and in particular
it allows me to test the PyInstaller bundle functionality on Linux, even
though we don't officially distribute it as such.
|
|
pyinstaller considers .py files to not be data by default. The entire
architecture of pyinstaller is, in fact, different from zipapp or
unpacked modules -- because it actually has to use a resource loader
with on-disk files *separate* from the module itself. So just because a
file gets packaged in the application does not mean it's usable as an
importlib.resources compatible resource.
Although we collect data files in general, we need to make sure that .py
files are collected from scripts, because we may try to run them
standalone from an external process.
Fixes #11689
|
|
This is the same fix as the one in #10800 for shared libraries but
applied to executables instead.
|
|
Running some container-like mechanisms such as chroot(1) from sudo, can
result in a new isolated environment where the environment variables
exist but no users exist. From there, a build is performed as root but
installation fails when we try to look up the passwd database entry for
the user outside of the chroot.
Proper container mechanisms such as systemd-nspawn, and even improper
ones like docker, sanitize this and ensure those stale environment
variables don't exist anymore. But chroot is very low-level.
Avoid crashing when this happens.
Fixes #11662
|
|
A user might run `sudo somewrapper` to build and install something with
meson, and it is not actually possible to drop privileges and build,
since the build directory is also owned by root.
A common case of this is `sudo pip install` for projects using
meson-python or other python build-backends that wrap around meson.
Fixes #11665
|
|
The paths in meson.build use / as path separator, however, the paths
constructed during the directory structure walk use native path
separators, thus the path never compare equal to the excluded ones.
Normalize the exclusion paths before the comparison.
|
|
name string
|
|
language
staticlib/cdylib only provide a C ABI, i.e. contain no Rust metadata,
and can be linked like a link target from any other language.
|
|
cdylibs provide a plain C ABI to its consumers and should not be treated
like dylib/proc-macro shared libraries that provide a Rust ABI.
|
|
Rust compiler
By default clippy-driver will report its own version, which is not very
useful to check the toolchain version. Instead make sure to extract the
actual toolchain version here.
|
|
Dashes can't be used as identifiers in Rust and cargo is converting
dashes in crate names to underscores for the same reason.
|
|
Rust doesn't have a concept of dependency compile arguments, i.e.
something like headers. Dependencies are linked in and all required
metadata is provided by the linker flags.
|
|
https://learn.microsoft.com/en-us/visualstudio/msbuild/cl-task?view=vs-2022
Found by https://www.check-spelling.dev/
Signed-off-by: Josh Soref <2119212+jsoref@users.noreply.github.com>
|
|
* --load-average
* --gdb-path
* description for install
Found by https://www.check-spelling.dev/
Signed-off-by: Josh Soref <2119212+jsoref@users.noreply.github.com>
|
|
Found by https://www.check-spelling.dev/
Signed-off-by: Josh Soref <2119212+jsoref@users.noreply.github.com>
|
|
Found by https://www.check-spelling.dev/
Signed-off-by: Josh Soref <2119212+jsoref@users.noreply.github.com>
|
|
Fixes regression from 388fa6e7761712a7b1d3422a879fa9e90b6fa606
Found by https://www.check-spelling.dev/
Signed-off-by: Josh Soref <2119212+jsoref@users.noreply.github.com>
|
|
This is used by the meson.pkg but is an optional attribute:
https://developer.apple.com/library/archive/documentation/DeveloperTools/Reference/DistributionDefinitionRef/Chapters/Distribution_XML_Ref.html#//apple_ref/doc/uid/TP40005370-CH100-SW41
Found by https://www.check-spelling.dev/
Signed-off-by: Josh Soref <2119212+jsoref@users.noreply.github.com>
|
|
Signed-off-by: Josh Soref <2119212+jsoref@users.noreply.github.com>
|
|
Instead of using a hacky version of
rpaths_for_non_system_absolute_shared_libraries,
we use a custom adhoc function. The function filters out
paths that do not contain dll files. It also finds paths
from any kind of dependencies.
This removes unneeded paths from the PATH env var,
and adds paths that were missing otherwise.
|
|
This reverts commit 3968b84696a9034233c2fd64ef7d8e078982afe3.
|
|
The word "cannot" expresses inability to do something whereas "can not"
expresses the ability to refrain from doing something.
|
|
|
|
|
|
|
|
|
|
During evaluation of codeblocks, we start off with an iteration of
nodes, and then while evaluating them we may update the global
self.current_node context. When catching and formatting errors, we
didn't take into account that the node might be updated from the
original top-level iteration.
Switch to formatting errors using self.current_node instead, to ensure
we can point at the likely most-accurate actual cause of an error.
Also update the current node in a few more places, so that function
calls always see the function call as the current node, even if the most
recently parsed node was an argument to the function call.
Fixes #11643
|
|
Use collapsible sections for cicmds.
|
|
It fails with "incompatible-function-pointer-types" otherwise
|
|
|
|
|
|
In commit 8da060706c2c34fbea91898509123cb190894f3e we fixed an issue
where stashing would ignore untracked files, potentially conflicting
with what we reset to.
Unfortunately in the process it broke the functionality entirely, by
producing an invalid git command... according to some versions of git.
This is actually a git bug, fixed upstream in git 2.37.0 via commit
b02fdbc80a41f73ceb6c99e8e27b22285243a335. It causes git itself to
re-synthesize the command line for a negation pathspec with a "."
rather than "", as that's what is needed to make it match all files
other than the negation.
For backwards compatibility to older versions of git, we manually pass
the dot pathspec ourselves.
|
|
Previously, the VS backend would add the the include directories
in the reverse order that the Ninja backend does it, which
means that the source directories would be preferred over
the build directories. For example, this would cause the
compiler to choose the file from the source directory
if a file with the same name is found in both.
Fixes 57ec097b5 ("vs: Use CompilerArgs() for compile and link args")
Fixes #11630
|
|
Extend the "common/include order" test to ensure that the build
directory is preferred over the source directory. For example,
when using `configure_file()`, the resulting file should be
preferred over a file with the same name in the source directory.
|
