To be a bit more constructive with this: The switch to C++11 should mean that the default boost build will be compiled in C++11+ (this looks strange, what I mean is: Use no compiler switch as before, unless the compiler would default to C++03 in which case pass -std=c++11) The reason for this is quite simple: ABI compatibility. This is a major bummer as you cannot e.g. mix std::string in C++03/11. Most projects (I assume) now start with at least C++11 and I think you can in most cases mix C++11 and C++14 (and maybe even more), at least more easily than C++03 with anything newer. And lastly: Yes you can #ifdef for C++11 features. But that increases maintenance costs. With C++11 and up as the minimum to support, such things will be a lot less. But I also want to highlight that "we" (some of the projects I'm working on) are using Boost BECAUSE of its C++03 compatibility. If some don't want or cannot switch to C++11 and up they can rely on Boost bringing a lot of functionality in. I've seen a lot of codes with something like `namespace myStd{ #if C++11 using namespace std; #else using namespace boost }` which did work quite well to support both standards downstream or midstream (libraries using boost) Having said that: 2020 sounds like a reasonable date. Enough time for preparation and migration. - Alex PS: As HPC Systems came up: They usually count in decades. GCC 4.8 was released in 2013, so rule of thumb would be that at about 2023 everyone is using GCC 4.8+. But you can also look for RHEL support (https://access.redhat.com/solutions/19458) RHEL 7 uses GCC 4.8 where RHEL 6 is supported till end of 2020 (not counting extended support)