Source: dune-common Version: 2.10.0-4 Severity: serious Tags: ftbfs patch User: [email protected] Usertags: ieee-long-double
Hi Maintainer We are working on https://wiki.debian.org/ToolChain/IEEELongDouble The attached patch was applied in Ubuntu after some discussion with upstream. Regards Graham
From: Florent 'Skia' Jacquet <[email protected]> Date: Fri, 20 Jun 2025 11:15:26 +0200 Subject: [PATCH] disable _GLIBCXX_USE_FLOAT128 for QuadMath Since recently, gcc-14 is shipped with --with-long-double-format=ieee for ppc64el. Having _GLIBCXX_USE_FLOAT128 defined breaks the build on this architecture, and it's not needed anymore to have working float128 anyway. Bug-Ubuntu: https://bugs.launchpad.net/ubuntu/+source/dune-grid/+bug/2114917 Forwarded: discussed on upstream's Matrix, and drafted by them here: https://gitlab.dune-project.org/core/dune-common/-/merge_requests/1552 --- cmake/modules/FindQuadMath.cmake | 3 --- 1 file changed, 3 deletions(-) --- a/cmake/modules/FindQuadMath.cmake +++ b/cmake/modules/FindQuadMath.cmake @@ -71,11 +71,4 @@ # Compiler supports QuadMath: Add appropriate linker flag add_library(QuadMath::QuadMath INTERFACE IMPORTED) target_link_libraries(QuadMath::QuadMath INTERFACE quadmath) - - target_compile_definitions(QuadMath::QuadMath INTERFACE - _GLIBCXX_USE_FLOAT128 - ) - target_compile_options(QuadMath::QuadMath INTERFACE - $<$<CXX_COMPILER_ID:GNU>:-fext-numeric-literals> - ) endif()
