That's an interesting point regarding changing idioms/culture.
Can't speak about Haskell, however I believe that C++ changes can be incrementally adopted: we had C++98, then boost established itself, some parts of boost were included in TR1 and then were part of C++11. A similar process is taking place with other libraries such as asio and filesystem. I don't have the feeling that modern C++ style is in flux, its pillars continue to be RAII, generic programming and the STL.
I think Bjarne & co have explicitly stated that they want to grow the language while maintaining backwards compatibility as much as possible and from my experience, the code I write today builds on patterns and idioms that started to crystalize a long time ago.
Yep RAII was already a thing in the late 90's for example.
Re:Two Weeks of Rust
by nickpsecurity on Tuesday January 12, 2016 @02:46PM
(#10889821)
That is an interesting point. I'd like to see pcwalton respond to it. That people are having problems with current Rust idioms means they might [rightly] change them. So, starting with Rust now could reinforce bad practices that come with a retraining or software re-engineering cost later. So, this is a real risk that I overlooked in prior discussions.
Good catch.
I'm myself willing to accept the promise of code stability, so I think you're looking at slightly the wrong problem. While I'd expect current code to continue to compile, I also expect what is regarded as good Rust now won't be in five years. That's not a bad thing, but it also means that hopping on the Rust bandwagon might be premature.
The worst case? How about Haskell, where new language features make old code look worse even as it still works? Or C++, with it's accelerating rate of style changes?