Re: firefox still fails on 2020Q1, netbsd-8, amd64

[Tobias Nygren <tnn%NetBSD.org@localhost>]

On Mon, 11 May 2020 14:42:41 +0100
Mike Pumford <mpumford%mudcovered.org.uk@localhost> wrote:

> On 11/05/2020 13:32, Patrick Welche wrote:
> 
> > I think
> > 
> > template< class T >
> > inline constexpr bool is_empty_v = is_empty<T>::value;
> > 
> > but only since c++17
> > 
> Just remembered something. clang comes with its own C++ runtime. Any 
> reason we are using clang with the gcc runtime rather than using the 
> clang one?
> 
> I know it has one as when you debug a clang c++ binary on gdb with the 
> c++ debug extensions to display STL stuff sanely those extensions fail 
> on clang compiled code due to different internals.
> 
> Is that a way we could get clang an up to date runtime for firefox 
> without having to do wierd stuff with gcc-libs?

We have this packaged in lang/libcxx. But if Firefox depends on other
libs (say, textproc/icu) that link with libstdc++.so rather than
libc++.so that will end bad as you have two incompatible implementations.