I don't see how it can affect the build as it's a TEST_DEPENDS only; I don't think there's any need to test building other ports, your "make test" should be enough.

It appears to just want any version of libclang, and when run normally via the script in bin/ it searches for it in /usr/local/llvm*/lib to set LIBCLANG_PATH before running the binary in libexec.

OK.

--
 Sent from a phone, apologies for poor formatting.

On 18 February 2025 07:16:03 Theo Buehler <t...@theobuehler.org> wrote:

Apparently this is the last consumer of llvm/17 and it doesn't seem to
need it. On amd64 all tests pass with the diff below, the same test
fails on aarch64 with llvm 17 and llvm 18, so I don't think this is
a blocker.

Not sure if this change requires a test build of one of the chromiums.
I don't currently have access to a beefy enough machine to do this myself.

test test_wrap_static_fns ... FAILED

failures:

---- test_wrap_static_fns stdout ----
In path is ::: tests/expectations/tests/generated/wrap_static_fns
Out path is ::: /usr/ports/pobj/rust-bindgen-0.69.4/build-aarch64/target/release/build/bindgen-tests-7829d1aaa511ddb4/out/wrap_static_fns
thread 'test_wrap_static_fns' panicked at bindgen-tests/tests/tests.rs:660:10:
Failed to generate bindings: Codegen(Serialize { msg: "Cannot serialize type kind Opaque", loc: "tests/headers/wrap-static-fns.h:63:56" })
stack backtrace:
0: 0x1247cb1210 - <std::sys::backtrace::BacktraceLock::print::DisplayBacktrace as core::fmt::Display>::fmt::hf547466864cc70c9
  1:       0x1247cee7dc - core::fmt::write::hcaa7796db5aa9bca
  2:       0x1247cd28cc - std::io::Write::write_fmt::h1633e398a9373372
3: 0x1247cb1104 - std::sys::backtrace::BacktraceLock::print::h4106fc81705c7a8d 4: 0x1247cc1744 - std::panicking::default_hook::{{closure}}::h8f52df955cd8c13b
  5:       0x1247cc1544 - std::panicking::default_hook::haf911d019667e488
  6:       0x12479e3d50 - test::test_main::{{closure}}::h3f7dfbefc34e5786
  7:       0x1247cc1d70 - 
std::panicking::rust_panic_with_hook::h32488bea288c74dc
8: 0x1247cb164c - std::panicking::begin_panic_handler::{{closure}}::h73a4b59a6837c893 9: 0x1247cb1474 - std::sys::backtrace::__rust_end_short_backtrace::hf629708e8567acf5
 10:       0x1247cc185c - rust_begin_unwind
 11:       0x1247cf3ca4 - core::panicking::panic_fmt::h09632188a6b3ef41
 12:       0x1247cf3ffc - core::result::unwrap_failed::h508c56a2a722bd66
13: 0x1247934278 - core::ops::function::FnOnce::call_once::h88f52abb689a8bb9
 14:       0x12479e87bc - test::__rust_begin_short_backtrace::ha01923d4b9b30ebc
 15:       0x12479e894c - test::run_test_in_process::h290566f3ab970e06
16: 0x1247a0d560 - std::sys::backtrace::__rust_begin_short_backtrace::hf25750de98ba0394 17: 0x12479f0d88 - core::ops::function::FnOnce::call_once{{vtable.shim}}::h62ac8f60fa927bb5 18: 0x1247cb9004 - std::sys::pal::unix::thread::Thread::new::thread_start::h7e62f47629a3a067
 19:       0x165f2c8eb0 - _rthread_start
                              at /usr/src/lib/librthread/rthread.c:96:11
 20:       0x16bf216864 - __tfork_thread
                              at 
/usr/src/lib/libc/arch/aarch64/sys/tfork_thread.S:43


Here's the diff. And yes, it needs to be MAKE_ENV, not TEST_ENV.

Index: Makefile
===================================================================
RCS file: /cvs/ports/devel/rust-bindgen/Makefile,v
diff -u -p -r1.1.1.1 Makefile
--- Makefile 13 Jun 2024 16:07:42 -0000 1.1.1.1
+++ Makefile 18 Feb 2025 07:10:07 -0000
@@ -3,6 +3,7 @@ COMMENT = automatically generates Rust F
GH_ACCOUNT = rust-lang
GH_PROJECT = rust-bindgen
GH_TAGNAME = v0.69.4
+REVISION = 0

CATEGORIES = devel

@@ -20,10 +21,10 @@ MODCARGO_INSTALL_TARGET_PATHS = bindgen-
CONFIGURE_STYLE = cargo
SEPARATE_BUILD = Yes

-TEST_DEPENDS += devel/llvm/17
+TEST_DEPENDS += devel/llvm/18

# for tests
-MAKE_ENV += LIBCLANG_PATH=${LOCALBASE}/llvm17/lib
+MAKE_ENV += LIBCLANG_PATH=${LOCALBASE}/llvm18/lib

post-install:
 mv ${PREFIX}/bin/bindgen ${PREFIX}/libexec/rust-bindgen

Reply via email to