pawosm-arm wrote: > I don't think it is correct to say that the option was ignored. It may still > be read by the compiler driver (which generates the link line). In fact it > probably _should_ be processed by the compiler driver in order to set the C++ > standard library correctly - otherwise you could just not set this in your > `LDFLAGS`.
But this (ignoring) is exactly that is happening. Contrary to classic flang, flang-new cannot be used to mix C++ and Fortran code in one compiler invocation, and working on this patch I've even found a test case guarding that controversial decision (`driver-error-cc1.cpp`). This is something that would probably trigger further discussion as angry users may soon emerge as flang-new's popularity grows, but for now, I've found that even if I have a project for which its Makefile builds C++ and Fortran sources separately, it fails at link time if I'm trying to use specific standard C++ library by setting designated CMake flag (which in turn, sets C++ and linker flags for the project globally). I could go through all CMake generated link.txt files and adjust the flags, but this exposes a terrible user experience as: 1) the user must be experienced enough to know how this must be modified 2) it introduces annoying extra step between cmake invocation and make/ninja invocation. Using this flag for compilation of a Fortran code will display a warining: ``` flang-new: warning: argument unused during compilation: '-stdlib=platform' [-Wunused-command-line-argument] ``` ...and this is expected. Using this flag for linking will not do anything, looking at `-###` output with and without using this flag shows no difference, ergo it is being ingored. https://github.com/llvm/llvm-project/pull/110598 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
