rmaprath added a comment. In http://reviews.llvm.org/D20328#456241, @mclow.lists wrote:
> > Apologies, not at program launch time but link time. > > > I'm OK with that; I think that's unnecessary complication, but not a > deal-breaker. > The choosing at program launch seems unworkable to me. > > [ I think that the threading implementation should be chosen by the provider > of libc++, frankly. ] The problem with the latter is the number of different platforms we have to support, each with different threading primitives. For most RTOS platforms, pthreads is a big ask and even if they can provide a subset of `<pthread>`, different platforms will have different definitions of `PTHREAD_MUTEX_INITIALIZER` for example, making it very difficult for us to build `libc++` in a platform agnostic way. http://reviews.llvm.org/D20328 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
