Sigh ... this test still isn't right, because it fails on systems without a bug-fix for TBB 2020.3. We could just add -fpermissive to the test options, but let's just require tbb_backend (which isn't matched on systems with the buggy headers). Separately, I'm going to backport the TBB fix to Fedora 37 where I had the test FAIL.
Tested x86_64-linux (Fedora 37 this time). Pushed to trunk. -- >8 -- The https://github.com/oneapi-src/oneTBB/pull/833 fix is needed for TBB headers to avoid an error with GCC 13. The new names_pstl.cc test needs to check effective target tbb_backend so that it's UNSUPPORTED on systems without fixed headers. libstdc++-v3/ChangeLog: * testsuite/17_intro/names_pstl.cc: Require et tbb_backend. --- libstdc++-v3/testsuite/17_intro/names_pstl.cc | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/libstdc++-v3/testsuite/17_intro/names_pstl.cc b/libstdc++-v3/testsuite/17_intro/names_pstl.cc index 42ce098550c..019e4774ff2 100644 --- a/libstdc++-v3/testsuite/17_intro/names_pstl.cc +++ b/libstdc++-v3/testsuite/17_intro/names_pstl.cc @@ -1,10 +1,12 @@ // { dg-do compile { target c++17 } } -// { dg-additional-options "-DTBB_SUPPRESS_DEPRECATED_MESSAGES=1" { target tbb_backend } } +// { dg-require-effective-target tbb_backend } +// { dg-add-options no_pch } // The TBB headers use non-reserved names (because they're not part of the // implementation) so we need to include them before the macro definitions // in names.cc: #if __has_include(<tbb/tbb.h>) +# define TBB_SUPPRESS_DEPRECATED_MESSAGES 1 # include <tbb/tbb.h> #endif // Now we can define the macros to poison uses of non-reserved names: -- 2.39.1