For me it's not about the language. It's about pragmatics. With C++ I can write code that builds natively using standard tool chains and libraries on Linux, Ma

For me it's not about the language. It's about pragmatics. With C++ I can write code that builds natively using standard tool chains and libraries on Linux, Mac, Windows, iOS, Android, and dozens of other less common platforms. With C++ I can write something once and then ship it on all those things without worrying (much) about the availability of tooling. If I write it using portable programming techniques (mostly acquired through hard experience) I will not have to do very much porting, and the #ifdef mechanism provides an acceptable way to accomplish alternate code paths on different platforms without compromising cleanliness of the build or performance. C++ has its uglies, but it's not too bad if one knows it well and avoids various also-learned-through-experience antipatterns. Go is in fact better than C++, but it's not better enough to justify the hassle or the impact in my ability to ship across umpteen platforms. This is why everyone still does systems programming in C/C++. Write (a bit painfully) once, ship everywhere. Other languages like Go, Ruby, Python, Rust, server-side JavaScript, various functional dialects, D, whatever, always achieve success in two areas: specific ecosystems/platforms (e.g. ObjC on Apple, Haskell in the world of high-speed algorithmic trading), and server-side coding. With server-side coding who cares if you can ship, cause you don't. But all the world is not a server, and sometimes you want to hit multiple user bases. For that, only C will do. I actually think a Go (or any other language) compiler that builds to cross-platform C code with a configure script and a Windows build script would be a worthy development. I know many people hate language-to-language compilers, but this would enable one to write in nifty new language X, build intermediate code, then build that intermediate code with the native tooling of every other damn thing. The other alternative would be for Google -- if it really wants to push Go -- to put a lot of effort behind porting it to offer good high-quality native tooling for everything and everyone's toaster oven. But Google thinks all the world's a server or a browser (and it is for Google for the most part) so that doesn't seem to be a priority for them.