I haven't looked at the patch yet and my knowledge of spec files is pretty weak, so sorry if these are stupid questions ...
On 7 June 2013 16:45, Rainer Orth wrote: > > Follow libgfortran's lead and introduce libstdc++.spec. Here, > target-specific linker switches can be determined by the libstdc++ > configury, with configuration well localized, and one can easily handle > stuff like linking -lrt with -z ignore/-z record > resp. --as-needed/--no-as-needed. This provides us with a general > extension mechanism for linking C++ programs, which can be used for > other purposes if the need arises. Could this later be extended to serve as the basis for a -cxxthread switch that does platform-specific things to allow the C++11 thread library to be used? I don't like that users must use -pthread for std::thread and std::mutex to work, because they shouldn't need to know or care whether it uses Pthreads internally, or some other thread model. > The downside of course is that the testsuite needs to be taught about > that file by passing the appropriate -B switches to find it at compile > time. Is that only necessary for testing an uninstalled libstdc++? In other words, would the spec file be installed and used automatically for an installed libstdc++? > One way or another, this needs to be adressed to fix the libstdc++ > testsuite failures on Solaris. The other option is to make --enable-libstdcxx-time=auto only effective for Solaris 11, but I think you said you'd prefer to keep it for 9 and 10.