The bug that is referenced in the SKIP message was merged in time for
the most recent llvm 16.x release. However, due to the graciousness of
the LLVM developers, a very reasonable response was taken:
LLVM will not merge ("the average"?) bugfixes during the final point
release of a release series, as judged by their willingness to continue
maintaining a major release of LLVM (????????) because merging a bugfix
could theoretically cause a new bug instead and that Simply Is Not
Done™. It could result in truly tragic outcomes, like having to release
another bugfix.
This innovative approach to bug solving has advanced the science of
computing forward by decades. Sadly, it comes at a downside: software
doesn't work. At this point it becomes obvious that llvm 16 in general
will simply not work with Meson, and this is "expected", so bump the
version checking for our SKIP to match reality.
In commit 89146e84c9 we added some
complicated code to verify the llvm framework's "combination" matrix
lookup. It expects to find llvm with both cmake and config-tool, with
the same version. But the sanity check is wonky -- it checks that both
have the same found status, instead, so if both are not found then we
proceed to try to convert the string "unknown" to a mapping of semver
integers, and this is guaranteed to fail.
This can happen for example if the system llvm exists in the general
case, but actual modules cannot be found because the system llvm does
not distribute static modules. For example, this is the case on Gentoo.
Abort more obviously by just insisting that both be found. If they
aren't both found, then investigative efforts know to look at why they
weren't found.
llvm-config is unsuitable for standard cross-compile,
because we need to build llvm especially for it, which
is not done is almost any distros, so, for example,
standard bootstrap chroot will be unsuitable.
This patch is trying to acheive feature parity between
config-tool searching of LLVM and CMake-based one,
which is arch-agnostic.
Signed-off-by: Konstantin <ria.freelander@gmail.com>
Remove hard-coded framework test skip logic in skippable(), instead
annotate test.json with environments in which skip is expected.
(Mainly this is done with by testing the value of MESON_CI_JOBNAME now
set for linux jobs)
If the required LLVM modules can't be found, skip the LLVM framework
test, rather than succesfully doing nothing.
(This optionality is a leftover from before #7379)
(At the moment, OpenSuse provides dynamic-only LLVM. The cmake method
still finds LLVM, when a static LLVM is requested, but fails to find any
modules. This might be a bug in the cmake method of the LLVM
dependency.)
* depenencies/llvm: Handle llvm-config --shared-mode failing
Fixes: #7371Fixes: #7878
* test cases/llvm: Refactor to use test.json
Instead of trying to cover everything internally
D lang compilers have an option -release (or similar) which turns off
asserts, contracts, and other runtime type checking. This patch wires
that up to the b_ndebug flag.
Fixes#7082
It turns out there's a bug in creating a sub dependency out of threads
in that we pass all of the kwargs from the parent to the
ThreadDependency instance. This demonstrates the bug.
I left a hack patch in a pull request for LLVM, and the result is that
LLVM doesn't link with static builds. The real problem was that some
distros have pkg-config for tinfo, other's don't, so the correct
solution is to use cpp_compiler.find_library if dependency() fails.
Archlinux doesn't package tinfo, and the non-distro maintained package
(in the AUR), doesn't provide a pkgconfig, it just symlinks libncurses.
So just pick something else. glib-2.0 is used elsewhere in meson's test
suite, so that should be safe.
This adds a depdendncy wrapper for llvm-config based on the wxwidgets
dependency. IT handles libs, version, include dir, and the llvm unique
concept of components. These components are individual pieces of the
LLVM library that may or may not be available depending on the platform.