On Fri, 03 Mar 2023 at 17:41:22 +0100, Sylvestre Ledru wrote: > Le 16/01/2023 à 16:44, Adrian Bunk a écrit : > > /usr/include/wasm32-wasi/c++/v1/__type_traits/is_array.h:40:22: error: > > reference to unresolved using declaration > > template <class _Tp, size_t _Np> struct _LIBCPP_TEMPLATE_VIS > > is_array<_Tp[_Np]> > > ^ > > /usr/include/wasm32-wasi/c++/v1/cstddef:52:1: note: using declaration > > annotated with 'using_if_exists' here > > using ::size_t _LIBCPP_USING_IF_EXISTS;
I've cut down what I think is the root cause of this compile failure to a more minimal bug report: see #1032317. I don't think this is *really* a regression, because in the version in bookworm, the autopkgtest didn't exercise compilation of C++ into WebAssembly at all. If I understand correctly, when compared with bookworm, the version in sid adds a new feature (the necessary -dev packages for compiling C++ into WebAssembly) and also adds a smoke-test for that feature, but the feature doesn't yet work as intended. Is that accurate? Did the new autopkgtest coverage pass on the maintainers' systems? I don't seen any sign of it having passed on ci.debian.net since 1:15.0.6-5~exp1 (see https://ci.debian.net/packages/l/llvm-toolchain-15/unstable/amd64/ which includes test results with all packages from unstable, and also test results with llvm-toolchain-15 taken from experimental and everything else from unstable). If you don't consider full C++ wasm support to be release-critical for bookworm, one option would be to remove the packages that enable it, together with their test coverage; that should get the autopkgtest passing again. However, I haven't looked at the diff between 15.0.6 and 15.0.7, so I don't know whether the release team would be likely to accept an unblock request or not. When enabling new features in experimental, please check the CI results before re-uploading to unstable. This can be done conveniently via <https://qa.debian.org/excuses.php?experimental=1&package=llvm-toolchain-15>, which asks the question: if the packages in unstable were all in testing, and the packages in experimental were all in unstable, would llvm-toolchain-15 be able to migrate? Thanks, smcv
